Vendor dependencies for 0.3.0 release

vendor/tracing-subscriber/src/field/debug.rs

@@ -0,0 +1,111 @@
+//! `MakeVisitor` wrappers for working with `fmt::Debug` fields.
+use super::{MakeVisitor, VisitFmt, VisitOutput};
+use tracing_core::field::{Field, Visit};
+
+use core::fmt;
+
+/// A visitor wrapper that ensures any `fmt::Debug` fields are formatted using
+/// the alternate (`:#`) formatter.
+#[derive(Debug, Clone)]
+pub struct Alt<V>(V);
+
+// TODO(eliza): When `error` as a primitive type is stable, add a
+// `DisplayErrors` wrapper...
+
+// === impl Alt ===
+//
+impl<V> Alt<V> {
+    /// Wraps the provided visitor so that any `fmt::Debug` fields are formatted
+    /// using the alternative (`:#`) formatter.
+    pub fn new(inner: V) -> Self {
+        Alt(inner)
+    }
+}
+
+impl<T, V> MakeVisitor<T> for Alt<V>
+where
+    V: MakeVisitor<T>,
+{
+    type Visitor = Alt<V::Visitor>;
+
+    #[inline]
+    fn make_visitor(&self, target: T) -> Self::Visitor {
+        Alt(self.0.make_visitor(target))
+    }
+}
+
+impl<V> Visit for Alt<V>
+where
+    V: Visit,
+{
+    #[inline]
+    fn record_f64(&mut self, field: &Field, value: f64) {
+        self.0.record_f64(field, value)
+    }
+
+    #[inline]
+    fn record_i64(&mut self, field: &Field, value: i64) {
+        self.0.record_i64(field, value)
+    }
+
+    #[inline]
+    fn record_u64(&mut self, field: &Field, value: u64) {
+        self.0.record_u64(field, value)
+    }
+
+    #[inline]
+    fn record_bool(&mut self, field: &Field, value: bool) {
+        self.0.record_bool(field, value)
+    }
+
+    /// Visit a string value.
+    fn record_str(&mut self, field: &Field, value: &str) {
+        self.0.record_str(field, value)
+    }
+
+    // TODO(eliza): add RecordError when stable
+    // fn record_error(&mut self, field: &Field, value: &(dyn std::error::Error + 'static)) {
+    //     self.record_debug(field, &format_args!("{}", value))
+    // }
+
+    #[inline]
+    fn record_debug(&mut self, field: &Field, value: &dyn fmt::Debug) {
+        self.0.record_debug(field, &format_args!("{:#?}", value))
+    }
+}
+
+impl<V, O> VisitOutput<O> for Alt<V>
+where
+    V: VisitOutput<O>,
+{
+    #[inline]
+    fn finish(self) -> O {
+        self.0.finish()
+    }
+}
+
+feature! {
+    #![feature = "std"]
+    use super::VisitWrite;
+    use std::io;
+
+    impl<V> VisitWrite for Alt<V>
+    where
+        V: VisitWrite,
+    {
+        #[inline]
+        fn writer(&mut self) -> &mut dyn io::Write {
+            self.0.writer()
+        }
+    }
+}
+
+impl<V> VisitFmt for Alt<V>
+where
+    V: VisitFmt,
+{
+    #[inline]
+    fn writer(&mut self) -> &mut dyn fmt::Write {
+        self.0.writer()
+    }
+}

vendor/tracing-subscriber/src/field/delimited.rs

@@ -0,0 +1,184 @@
+//! A `MakeVisitor` wrapper that separates formatted fields with a delimiter.
+use super::{MakeVisitor, VisitFmt, VisitOutput};
+
+use core::fmt;
+use tracing_core::field::{Field, Visit};
+
+/// A `MakeVisitor` wrapper that wraps a visitor that writes formatted output so
+/// that a delimiter is inserted between writing formatted field values.
+#[derive(Debug, Clone)]
+pub struct Delimited<D, V> {
+    delimiter: D,
+    inner: V,
+}
+
+/// A visitor wrapper that inserts a delimiter after the wrapped visitor formats
+/// a field value.
+#[derive(Debug)]
+pub struct VisitDelimited<D, V> {
+    delimiter: D,
+    seen: bool,
+    inner: V,
+    err: fmt::Result,
+}
+
+// === impl Delimited ===
+
+impl<D, V, T> MakeVisitor<T> for Delimited<D, V>
+where
+    D: AsRef<str> + Clone,
+    V: MakeVisitor<T>,
+    V::Visitor: VisitFmt,
+{
+    type Visitor = VisitDelimited<D, V::Visitor>;
+    fn make_visitor(&self, target: T) -> Self::Visitor {
+        let inner = self.inner.make_visitor(target);
+        VisitDelimited::new(self.delimiter.clone(), inner)
+    }
+}
+
+impl<D, V> Delimited<D, V> {
+    /// Returns a new [`MakeVisitor`] implementation that wraps `inner` so that
+    /// it will format each visited field separated by the provided `delimiter`.
+    ///
+    /// [`MakeVisitor`]: super::MakeVisitor
+    pub fn new(delimiter: D, inner: V) -> Self {
+        Self { delimiter, inner }
+    }
+}
+
+// === impl VisitDelimited ===
+
+impl<D, V> VisitDelimited<D, V> {
+    /// Returns a new [`Visit`] implementation that wraps `inner` so that
+    /// each formatted field is separated by the provided `delimiter`.
+    ///
+    /// [`Visit`]: tracing_core::field::Visit
+    pub fn new(delimiter: D, inner: V) -> Self {
+        Self {
+            delimiter,
+            inner,
+            seen: false,
+            err: Ok(()),
+        }
+    }
+
+    fn delimit(&mut self)
+    where
+        V: VisitFmt,
+        D: AsRef<str>,
+    {
+        if self.err.is_err() {
+            return;
+        }
+
+        if self.seen {
+            self.err = self.inner.writer().write_str(self.delimiter.as_ref());
+        }
+
+        self.seen = true;
+    }
+}
+
+impl<D, V> Visit for VisitDelimited<D, V>
+where
+    V: VisitFmt,
+    D: AsRef<str>,
+{
+    fn record_i64(&mut self, field: &Field, value: i64) {
+        self.delimit();
+        self.inner.record_i64(field, value);
+    }
+
+    fn record_u64(&mut self, field: &Field, value: u64) {
+        self.delimit();
+        self.inner.record_u64(field, value);
+    }
+
+    fn record_bool(&mut self, field: &Field, value: bool) {
+        self.delimit();
+        self.inner.record_bool(field, value);
+    }
+
+    fn record_str(&mut self, field: &Field, value: &str) {
+        self.delimit();
+        self.inner.record_str(field, value);
+    }
+
+    fn record_debug(&mut self, field: &Field, value: &dyn fmt::Debug) {
+        self.delimit();
+        self.inner.record_debug(field, value);
+    }
+}
+
+impl<D, V> VisitOutput<fmt::Result> for VisitDelimited<D, V>
+where
+    V: VisitFmt,
+    D: AsRef<str>,
+{
+    fn finish(self) -> fmt::Result {
+        self.err?;
+        self.inner.finish()
+    }
+}
+
+impl<D, V> VisitFmt for VisitDelimited<D, V>
+where
+    V: VisitFmt,
+    D: AsRef<str>,
+{
+    fn writer(&mut self) -> &mut dyn fmt::Write {
+        self.inner.writer()
+    }
+}
+
+#[cfg(test)]
+#[cfg(all(test, feature = "alloc"))]
+mod test {
+    use super::*;
+    use crate::field::test_util::*;
+
+    #[test]
+    fn delimited_visitor() {
+        let mut s = String::new();
+        let visitor = DebugVisitor::new(&mut s);
+        let mut visitor = VisitDelimited::new(", ", visitor);
+
+        TestAttrs1::with(|attrs| attrs.record(&mut visitor));
+        visitor.finish().unwrap();
+
+        assert_eq!(
+            s.as_str(),
+            "question=\"life, the universe, and everything\", tricky=true, can_you_do_it=true"
+        );
+    }
+
+    #[test]
+    fn delimited_new_visitor() {
+        let make = Delimited::new("; ", MakeDebug);
+
+        TestAttrs1::with(|attrs| {
+            let mut s = String::new();
+            {
+                let mut v = make.make_visitor(&mut s);
+                attrs.record(&mut v);
+            }
+            assert_eq!(
+                s.as_str(),
+                "question=\"life, the universe, and everything\"; tricky=true; can_you_do_it=true"
+            );
+        });
+
+        TestAttrs2::with(|attrs| {
+            let mut s = String::new();
+            {
+                let mut v = make.make_visitor(&mut s);
+                attrs.record(&mut v);
+            }
+            assert_eq!(
+                s.as_str(),
+                "question=None; question.answer=42; tricky=true; can_you_do_it=false"
+            );
+        });
+    }
+}

vendor/tracing-subscriber/src/field/display.rs

@@ -0,0 +1,117 @@
+//! `MakeVisitor` wrappers for working with `fmt::Display` fields.
+use super::{MakeVisitor, VisitFmt, VisitOutput};
+use tracing_core::field::{Field, Visit};
+
+use core::fmt;
+
+/// A visitor wrapper that ensures any strings named "message" are formatted
+/// using `fmt::Display`
+#[derive(Debug, Clone)]
+pub struct Messages<V>(V);
+
+// TODO(eliza): When `error` as a primitive type is stable, add a
+// `DisplayErrors` wrapper...
+
+// === impl Messages ===
+//
+impl<V> Messages<V> {
+    /// Returns a new [`MakeVisitor`] implementation that will wrap `inner` so
+    /// that any strings named `message` are formatted using `fmt::Display`.
+    ///
+    /// [`MakeVisitor`]: super::MakeVisitor
+    pub fn new(inner: V) -> Self {
+        Messages(inner)
+    }
+}
+
+impl<T, V> MakeVisitor<T> for Messages<V>
+where
+    V: MakeVisitor<T>,
+{
+    type Visitor = Messages<V::Visitor>;
+
+    #[inline]
+    fn make_visitor(&self, target: T) -> Self::Visitor {
+        Messages(self.0.make_visitor(target))
+    }
+}
+
+impl<V> Visit for Messages<V>
+where
+    V: Visit,
+{
+    #[inline]
+    fn record_f64(&mut self, field: &Field, value: f64) {
+        self.0.record_f64(field, value)
+    }
+
+    #[inline]
+    fn record_i64(&mut self, field: &Field, value: i64) {
+        self.0.record_i64(field, value)
+    }
+
+    #[inline]
+    fn record_u64(&mut self, field: &Field, value: u64) {
+        self.0.record_u64(field, value)
+    }
+
+    #[inline]
+    fn record_bool(&mut self, field: &Field, value: bool) {
+        self.0.record_bool(field, value)
+    }
+
+    /// Visit a string value.
+    fn record_str(&mut self, field: &Field, value: &str) {
+        if field.name() == "message" {
+            self.0.record_debug(field, &format_args!("{}", value))
+        } else {
+            self.0.record_str(field, value)
+        }
+    }
+
+    // TODO(eliza): add RecordError when stable
+    // fn record_error(&mut self, field: &Field, value: &(dyn std::error::Error + 'static)) {
+    //     self.record_debug(field, &format_args!("{}", value))
+    // }
+
+    #[inline]
+    fn record_debug(&mut self, field: &Field, value: &dyn fmt::Debug) {
+        self.0.record_debug(field, value)
+    }
+}
+
+impl<V, O> VisitOutput<O> for Messages<V>
+where
+    V: VisitOutput<O>,
+{
+    #[inline]
+    fn finish(self) -> O {
+        self.0.finish()
+    }
+}
+
+feature! {
+    #![feature = "std"]
+    use super::VisitWrite;
+    use std::io;
+
+    impl<V> VisitWrite for Messages<V>
+    where
+        V: VisitWrite,
+    {
+        #[inline]
+        fn writer(&mut self) -> &mut dyn io::Write {
+            self.0.writer()
+        }
+    }
+}
+
+impl<V> VisitFmt for Messages<V>
+where
+    V: VisitFmt,
+{
+    #[inline]
+    fn writer(&mut self) -> &mut dyn fmt::Write {
+        self.0.writer()
+    }
+}

vendor/tracing-subscriber/src/field/mod.rs

@@ -0,0 +1,365 @@
+//! Utilities for working with [fields] and [field visitors].
+//!
+//! [fields]: tracing_core::field
+//! [field visitors]: tracing_core::field::Visit
+use core::{fmt, marker::PhantomData};
+pub use tracing_core::field::Visit;
+use tracing_core::{
+    span::{Attributes, Record},
+    Event,
+};
+pub mod debug;
+pub mod delimited;
+pub mod display;
+
+/// Creates new [visitors].
+///
+/// A type implementing `MakeVisitor` represents a composable factory for types
+/// implementing the [`Visit` trait][visitors]. The `MakeVisitor` trait defines
+/// a single function, `make_visitor`, which takes in a `T`-typed `target` and
+/// returns a type implementing `Visit` configured for that target. A target may
+/// be a string, output stream, or data structure that the visitor will record
+/// data to, configuration variables that determine the visitor's behavior, or
+/// `()` when no input is required to produce a visitor.
+///
+/// [visitors]: tracing_core::field::Visit
+pub trait MakeVisitor<T> {
+    /// The visitor type produced by this `MakeVisitor`.
+    type Visitor: Visit;
+
+    /// Make a new visitor for the provided `target`.
+    fn make_visitor(&self, target: T) -> Self::Visitor;
+}
+
+/// A [visitor] that produces output once it has visited a set of fields.
+///
+/// [visitor]: tracing_core::field::Visit
+pub trait VisitOutput<Out>: Visit {
+    /// Completes the visitor, returning any output.
+    ///
+    /// This is called once a full set of fields has been visited.
+    fn finish(self) -> Out;
+
+    /// Visit a set of fields, and return the output of finishing the visitor
+    /// once the fields have been visited.
+    fn visit<R>(mut self, fields: &R) -> Out
+    where
+        R: RecordFields,
+        Self: Sized,
+    {
+        fields.record(&mut self);
+        self.finish()
+    }
+}
+
+/// Extension trait implemented by types which can be recorded by a [visitor].
+///
+/// This allows writing code that is generic over `tracing_core`'s
+/// [`span::Attributes`][attr], [`span::Record`][rec], and [`Event`]
+/// types. These types all provide inherent `record` methods that allow a
+/// visitor to record their fields, but there is no common trait representing this.
+///
+/// With `RecordFields`, we can write code like this:
+/// ```
+/// use tracing_core::field::Visit;
+/// # use tracing_core::field::Field;
+/// use tracing_subscriber::field::RecordFields;
+///
+/// struct MyVisitor {
+///     // ...
+/// }
+/// # impl MyVisitor { fn new() -> Self { Self{} } }
+/// impl Visit for MyVisitor {
+///     // ...
+/// # fn record_debug(&mut self, _: &Field, _: &dyn std::fmt::Debug) {}
+/// }
+///
+/// fn record_with_my_visitor<R>(r: R)
+/// where
+///     R: RecordFields,
+/// {
+///     let mut visitor = MyVisitor::new();
+///     r.record(&mut visitor);
+/// }
+/// ```
+/// [visitor]: tracing_core::field::Visit
+/// [attr]: tracing_core::span::Attributes
+/// [rec]: tracing_core::span::Record
+pub trait RecordFields: crate::sealed::Sealed<RecordFieldsMarker> {
+    /// Record all the fields in `self` with the provided `visitor`.
+    fn record(&self, visitor: &mut dyn Visit);
+}
+
+/// Extension trait implemented for all `MakeVisitor` implementations that
+/// produce a visitor implementing `VisitOutput`.
+pub trait MakeOutput<T, Out>
+where
+    Self: MakeVisitor<T> + crate::sealed::Sealed<(T, Out)>,
+    Self::Visitor: VisitOutput<Out>,
+{
+    /// Visits all fields in `fields` with a new visitor constructed from
+    /// `target`.
+    fn visit_with<F>(&self, target: T, fields: &F) -> Out
+    where
+        F: RecordFields,
+    {
+        self.make_visitor(target).visit(fields)
+    }
+}
+
+feature! {
+    #![feature = "std"]
+    use std::io;
+
+    /// Extension trait implemented by visitors to indicate that they write to an
+    /// `io::Write` instance, and allow access to that writer.
+    pub trait VisitWrite: VisitOutput<Result<(), io::Error>> {
+        /// Returns the writer that this visitor writes to.
+        fn writer(&mut self) -> &mut dyn io::Write;
+    }
+}
+
+/// Extension trait implemented by visitors to indicate that they write to a
+/// `fmt::Write` instance, and allow access to that writer.
+pub trait VisitFmt: VisitOutput<fmt::Result> {
+    /// Returns the formatter that this visitor writes to.
+    fn writer(&mut self) -> &mut dyn fmt::Write;
+}
+
+/// Extension trait providing `MakeVisitor` combinators.
+pub trait MakeExt<T>
+where
+    Self: MakeVisitor<T> + Sized,
+    Self: crate::sealed::Sealed<MakeExtMarker<T>>,
+{
+    /// Wraps `self` so that any `fmt::Debug` fields are recorded using the
+    /// alternate formatter (`{:#?}`).
+    fn debug_alt(self) -> debug::Alt<Self> {
+        debug::Alt::new(self)
+    }
+
+    /// Wraps `self` so that any string fields named "message" are recorded
+    /// using `fmt::Display`.
+    fn display_messages(self) -> display::Messages<Self> {
+        display::Messages::new(self)
+    }
+
+    /// Wraps `self` so that when fields are formatted to a writer, they are
+    /// separated by the provided `delimiter`.
+    fn delimited<D>(self, delimiter: D) -> delimited::Delimited<D, Self>
+    where
+        D: AsRef<str> + Clone,
+        Self::Visitor: VisitFmt,
+    {
+        delimited::Delimited::new(delimiter, self)
+    }
+}
+
+// === impl RecordFields ===
+
+impl crate::sealed::Sealed<RecordFieldsMarker> for Event<'_> {}
+impl RecordFields for Event<'_> {
+    fn record(&self, visitor: &mut dyn Visit) {
+        Event::record(self, visitor)
+    }
+}
+
+impl crate::sealed::Sealed<RecordFieldsMarker> for Attributes<'_> {}
+impl RecordFields for Attributes<'_> {
+    fn record(&self, visitor: &mut dyn Visit) {
+        Attributes::record(self, visitor)
+    }
+}
+
+impl crate::sealed::Sealed<RecordFieldsMarker> for Record<'_> {}
+impl RecordFields for Record<'_> {
+    fn record(&self, visitor: &mut dyn Visit) {
+        Record::record(self, visitor)
+    }
+}
+
+impl<F> crate::sealed::Sealed<RecordFieldsMarker> for &F where F: RecordFields {}
+impl<F> RecordFields for &F
+where
+    F: RecordFields,
+{
+    fn record(&self, visitor: &mut dyn Visit) {
+        F::record(*self, visitor)
+    }
+}
+
+// === blanket impls ===
+
+impl<T, V, F> MakeVisitor<T> for F
+where
+    F: Fn(T) -> V,
+    V: Visit,
+{
+    type Visitor = V;
+    fn make_visitor(&self, target: T) -> Self::Visitor {
+        (self)(target)
+    }
+}
+
+impl<T, Out, M> crate::sealed::Sealed<(T, Out)> for M
+where
+    M: MakeVisitor<T>,
+    M::Visitor: VisitOutput<Out>,
+{
+}
+
+impl<T, Out, M> MakeOutput<T, Out> for M
+where
+    M: MakeVisitor<T>,
+    M::Visitor: VisitOutput<Out>,
+{
+}
+
+impl<T, M> crate::sealed::Sealed<MakeExtMarker<T>> for M where M: MakeVisitor<T> + Sized {}
+
+impl<T, M> MakeExt<T> for M
+where
+    M: MakeVisitor<T> + Sized,
+    M: crate::sealed::Sealed<MakeExtMarker<T>>,
+{
+}
+
+#[derive(Debug)]
+#[doc(hidden)]
+pub struct MakeExtMarker<T> {
+    _p: PhantomData<T>,
+}
+
+#[derive(Debug)]
+#[doc(hidden)]
+pub struct RecordFieldsMarker {
+    _p: (),
+}
+
+#[cfg(all(test, feature = "alloc"))]
+#[macro_use]
+pub(in crate::field) mod test_util {
+    use super::*;
+    pub(in crate::field) use alloc::string::String;
+    use tracing_core::{
+        callsite::Callsite,
+        field::{Field, Value},
+        metadata::{Kind, Level, Metadata},
+    };
+
+    pub(crate) struct TestAttrs1;
+    pub(crate) struct TestAttrs2;
+
+    impl TestAttrs1 {
+        pub(crate) fn with<T>(f: impl FnOnce(Attributes<'_>) -> T) -> T {
+            let fieldset = TEST_META_1.fields();
+            let values = &[
+                (
+                    &fieldset.field("question").unwrap(),
+                    Some(&"life, the universe, and everything" as &dyn Value),
+                ),
+                (&fieldset.field("question.answer").unwrap(), None),
+                (
+                    &fieldset.field("tricky").unwrap(),
+                    Some(&true as &dyn Value),
+                ),
+                (
+                    &fieldset.field("can_you_do_it").unwrap(),
+                    Some(&true as &dyn Value),
+                ),
+            ];
+            let valueset = fieldset.value_set(values);
+            let attrs = tracing_core::span::Attributes::new(&TEST_META_1, &valueset);
+            f(attrs)
+        }
+    }
+
+    impl TestAttrs2 {
+        pub(crate) fn with<T>(f: impl FnOnce(Attributes<'_>) -> T) -> T {
+            let fieldset = TEST_META_1.fields();
+            let none = tracing_core::field::debug(&Option::<&str>::None);
+            let values = &[
+                (
+                    &fieldset.field("question").unwrap(),
+                    Some(&none as &dyn Value),
+                ),
+                (
+                    &fieldset.field("question.answer").unwrap(),
+                    Some(&42 as &dyn Value),
+                ),
+                (
+                    &fieldset.field("tricky").unwrap(),
+                    Some(&true as &dyn Value),
+                ),
+                (
+                    &fieldset.field("can_you_do_it").unwrap(),
+                    Some(&false as &dyn Value),
+                ),
+            ];
+            let valueset = fieldset.value_set(values);
+            let attrs = tracing_core::span::Attributes::new(&TEST_META_1, &valueset);
+            f(attrs)
+        }
+    }
+
+    struct TestCallsite1;
+    static TEST_CALLSITE_1: &'static dyn Callsite = &TestCallsite1;
+    static TEST_META_1: Metadata<'static> = tracing_core::metadata! {
+        name: "field_test1",
+        target: module_path!(),
+        level: Level::INFO,
+        fields: &["question", "question.answer", "tricky", "can_you_do_it"],
+        callsite: TEST_CALLSITE_1,
+        kind: Kind::SPAN,
+    };
+
+    impl Callsite for TestCallsite1 {
+        fn set_interest(&self, _: tracing_core::subscriber::Interest) {
+            unimplemented!()
+        }
+
+        fn metadata(&self) -> &Metadata<'_> {
+            &TEST_META_1
+        }
+    }
+
+    pub(crate) struct MakeDebug;
+    pub(crate) struct DebugVisitor<'a> {
+        writer: &'a mut dyn fmt::Write,
+        err: fmt::Result,
+    }
+
+    impl<'a> DebugVisitor<'a> {
+        pub(crate) fn new(writer: &'a mut dyn fmt::Write) -> Self {
+            Self {
+                writer,
+                err: Ok(()),
+            }
+        }
+    }
+
+    impl Visit for DebugVisitor<'_> {
+        fn record_debug(&mut self, field: &Field, value: &dyn fmt::Debug) {
+            write!(self.writer, "{}={:?}", field, value).unwrap();
+        }
+    }
+
+    impl VisitOutput<fmt::Result> for DebugVisitor<'_> {
+        fn finish(self) -> fmt::Result {
+            self.err
+        }
+    }
+
+    impl VisitFmt for DebugVisitor<'_> {
+        fn writer(&mut self) -> &mut dyn fmt::Write {
+            self.writer
+        }
+    }
+
+    impl<'a> MakeVisitor<&'a mut dyn fmt::Write> for MakeDebug {
+        type Visitor = DebugVisitor<'a>;
+        fn make_visitor(&self, w: &'a mut dyn fmt::Write) -> DebugVisitor<'a> {
+            DebugVisitor::new(w)
+        }
+    }
+}

vendor/tracing-subscriber/src/filter/directive.rs

@@ -0,0 +1,457 @@
+use crate::filter::level::{self, LevelFilter};
+#[cfg(not(feature = "smallvec"))]
+use alloc::vec;
+#[cfg(not(feature = "std"))]
+use alloc::{string::String, vec::Vec};
+
+use core::{cmp::Ordering, fmt, iter::FromIterator, slice, str::FromStr};
+use tracing_core::{Level, Metadata};
+/// Indicates that a string could not be parsed as a filtering directive.
+#[derive(Debug)]
+pub struct ParseError {
+    kind: ParseErrorKind,
+}
+
+/// A directive which will statically enable or disable a given callsite.
+///
+/// Unlike a dynamic directive, this can be cached by the callsite.
+#[derive(Debug, PartialEq, Eq, Clone)]
+pub(crate) struct StaticDirective {
+    pub(in crate::filter) target: Option<String>,
+    pub(in crate::filter) field_names: Vec<String>,
+    pub(in crate::filter) level: LevelFilter,
+}
+
+#[cfg(feature = "smallvec")]
+pub(crate) type FilterVec<T> = smallvec::SmallVec<[T; 8]>;
+#[cfg(not(feature = "smallvec"))]
+pub(crate) type FilterVec<T> = Vec<T>;
+
+#[derive(Debug, PartialEq, Clone)]
+pub(in crate::filter) struct DirectiveSet<T> {
+    directives: FilterVec<T>,
+    pub(in crate::filter) max_level: LevelFilter,
+}
+
+pub(in crate::filter) trait Match {
+    fn cares_about(&self, meta: &Metadata<'_>) -> bool;
+    fn level(&self) -> &LevelFilter;
+}
+
+#[derive(Debug)]
+enum ParseErrorKind {
+    #[cfg(feature = "std")]
+    Field(Box<dyn std::error::Error + Send + Sync>),
+    Level(level::ParseError),
+    Other(Option<&'static str>),
+}
+
+// === impl DirectiveSet ===
+
+impl<T> DirectiveSet<T> {
+    // this is only used by `env-filter`.
+    #[cfg(all(feature = "std", feature = "env-filter"))]
+    pub(crate) fn is_empty(&self) -> bool {
+        self.directives.is_empty()
+    }
+
+    pub(crate) fn iter(&self) -> slice::Iter<'_, T> {
+        self.directives.iter()
+    }
+}
+
+impl<T: Ord> Default for DirectiveSet<T> {
+    fn default() -> Self {
+        Self {
+            directives: FilterVec::new(),
+            max_level: LevelFilter::OFF,
+        }
+    }
+}
+
+impl<T: Match + Ord> DirectiveSet<T> {
+    pub(crate) fn directives(&self) -> impl Iterator<Item = &T> {
+        self.directives.iter()
+    }
+
+    pub(crate) fn directives_for<'a>(
+        &'a self,
+        metadata: &'a Metadata<'a>,
+    ) -> impl Iterator<Item = &'a T> + 'a {
+        self.directives().filter(move |d| d.cares_about(metadata))
+    }
+
+    pub(crate) fn add(&mut self, directive: T) {
+        // does this directive enable a more verbose level than the current
+        // max? if so, update the max level.
+        let level = *directive.level();
+        if level > self.max_level {
+            self.max_level = level;
+        }
+        // insert the directive into the vec of directives, ordered by
+        // specificity (length of target + number of field filters). this
+        // ensures that, when finding a directive to match a span or event, we
+        // search the directive set in most specific first order.
+        match self.directives.binary_search(&directive) {
+            Ok(i) => self.directives[i] = directive,
+            Err(i) => self.directives.insert(i, directive),
+        }
+    }
+
+    #[cfg(test)]
+    pub(in crate::filter) fn into_vec(self) -> FilterVec<T> {
+        self.directives
+    }
+}
+
+impl<T: Match + Ord> FromIterator<T> for DirectiveSet<T> {
+    fn from_iter<I: IntoIterator<Item = T>>(iter: I) -> Self {
+        let mut this = Self::default();
+        this.extend(iter);
+        this
+    }
+}
+
+impl<T: Match + Ord> Extend<T> for DirectiveSet<T> {
+    fn extend<I: IntoIterator<Item = T>>(&mut self, iter: I) {
+        for directive in iter.into_iter() {
+            self.add(directive);
+        }
+    }
+}
+
+impl<T> IntoIterator for DirectiveSet<T> {
+    type Item = T;
+
+    #[cfg(feature = "smallvec")]
+    type IntoIter = smallvec::IntoIter<[T; 8]>;
+    #[cfg(not(feature = "smallvec"))]
+    type IntoIter = vec::IntoIter<T>;
+
+    fn into_iter(self) -> Self::IntoIter {
+        self.directives.into_iter()
+    }
+}
+
+// === impl Statics ===
+
+impl DirectiveSet<StaticDirective> {
+    pub(crate) fn enabled(&self, meta: &Metadata<'_>) -> bool {
+        let level = meta.level();
+        match self.directives_for(meta).next() {
+            Some(d) => d.level >= *level,
+            None => false,
+        }
+    }
+
+    /// Same as `enabled` above, but skips `Directive`'s with fields.
+    pub(crate) fn target_enabled(&self, target: &str, level: &Level) -> bool {
+        match self.directives_for_target(target).next() {
+            Some(d) => d.level >= *level,
+            None => false,
+        }
+    }
+
+    pub(crate) fn directives_for_target<'a>(
+        &'a self,
+        target: &'a str,
+    ) -> impl Iterator<Item = &'a StaticDirective> + 'a {
+        self.directives()
+            .filter(move |d| d.cares_about_target(target))
+    }
+}
+
+// === impl StaticDirective ===
+
+impl StaticDirective {
+    pub(in crate::filter) fn new(
+        target: Option<String>,
+        field_names: Vec<String>,
+        level: LevelFilter,
+    ) -> Self {
+        Self {
+            target,
+            field_names,
+            level,
+        }
+    }
+
+    pub(in crate::filter) fn cares_about_target(&self, to_check: &str) -> bool {
+        // Does this directive have a target filter, and does it match the
+        // metadata's target?
+        if let Some(ref target) = self.target {
+            if !to_check.starts_with(&target[..]) {
+                return false;
+            }
+        }
+
+        if !self.field_names.is_empty() {
+            return false;
+        }
+
+        true
+    }
+}
+
+impl Ord for StaticDirective {
+    fn cmp(&self, other: &StaticDirective) -> Ordering {
+        // We attempt to order directives by how "specific" they are. This
+        // ensures that we try the most specific directives first when
+        // attempting to match a piece of metadata.
+
+        // First, we compare based on whether a target is specified, and the
+        // lengths of those targets if both have targets.
+        let ordering = self
+            .target
+            .as_ref()
+            .map(String::len)
+            .cmp(&other.target.as_ref().map(String::len))
+            // Then we compare how many field names are matched by each directive.
+            .then_with(|| self.field_names.len().cmp(&other.field_names.len()))
+            // Finally, we fall back to lexicographical ordering if the directives are
+            // equally specific. Although this is no longer semantically important,
+            // we need to define a total ordering to determine the directive's place
+            // in the BTreeMap.
+            .then_with(|| {
+                self.target
+                    .cmp(&other.target)
+                    .then_with(|| self.field_names[..].cmp(&other.field_names[..]))
+            })
+            .reverse();
+
+        #[cfg(debug_assertions)]
+        {
+            if ordering == Ordering::Equal {
+                debug_assert_eq!(
+                    self.target, other.target,
+                    "invariant violated: Ordering::Equal must imply a.target == b.target"
+                );
+                debug_assert_eq!(
+                    self.field_names, other.field_names,
+                    "invariant violated: Ordering::Equal must imply a.field_names == b.field_names"
+                );
+            }
+        }
+
+        ordering
+    }
+}
+
+impl PartialOrd for StaticDirective {
+    fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
+        Some(self.cmp(other))
+    }
+}
+
+impl Match for StaticDirective {
+    fn cares_about(&self, meta: &Metadata<'_>) -> bool {
+        // Does this directive have a target filter, and does it match the
+        // metadata's target?
+        if let Some(ref target) = self.target {
+            if !meta.target().starts_with(&target[..]) {
+                return false;
+            }
+        }
+
+        if meta.is_event() && !self.field_names.is_empty() {
+            let fields = meta.fields();
+            for name in &self.field_names {
+                if fields.field(name).is_none() {
+                    return false;
+                }
+            }
+        }
+
+        true
+    }
+
+    fn level(&self) -> &LevelFilter {
+        &self.level
+    }
+}
+
+impl Default for StaticDirective {
+    fn default() -> Self {
+        StaticDirective {
+            target: None,
+            field_names: Vec::new(),
+            level: LevelFilter::ERROR,
+        }
+    }
+}
+
+impl fmt::Display for StaticDirective {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        let mut wrote_any = false;
+        if let Some(ref target) = self.target {
+            fmt::Display::fmt(target, f)?;
+            wrote_any = true;
+        }
+
+        if !self.field_names.is_empty() {
+            f.write_str("[")?;
+
+            let mut fields = self.field_names.iter();
+            if let Some(field) = fields.next() {
+                write!(f, "{{{}", field)?;
+                for field in fields {
+                    write!(f, ",{}", field)?;
+                }
+                f.write_str("}")?;
+            }
+
+            f.write_str("]")?;
+            wrote_any = true;
+        }
+
+        if wrote_any {
+            f.write_str("=")?;
+        }
+
+        fmt::Display::fmt(&self.level, f)
+    }
+}
+
+impl FromStr for StaticDirective {
+    type Err = ParseError;
+
+    fn from_str(s: &str) -> Result<Self, Self::Err> {
+        // This method parses a filtering directive in one of the following
+        // forms:
+        //
+        // * `foo=trace` (TARGET=LEVEL)
+        // * `foo[{bar,baz}]=info` (TARGET[{FIELD,+}]=LEVEL)
+        // * `trace` (bare LEVEL)
+        // * `foo` (bare TARGET)
+        let mut split = s.split('=');
+        let part0 = split
+            .next()
+            .ok_or_else(|| ParseError::msg("string must not be empty"))?;
+
+        // Directive includes an `=`:
+        // * `foo=trace`
+        // * `foo[{bar}]=trace`
+        // * `foo[{bar,baz}]=trace`
+        if let Some(part1) = split.next() {
+            if split.next().is_some() {
+                return Err(ParseError::msg(
+                    "too many '=' in filter directive, expected 0 or 1",
+                ));
+            }
+
+            let mut split = part0.split("[{");
+            let target = split.next().map(String::from);
+            let mut field_names = Vec::new();
+            // Directive includes fields:
+            // * `foo[{bar}]=trace`
+            // * `foo[{bar,baz}]=trace`
+            if let Some(maybe_fields) = split.next() {
+                if split.next().is_some() {
+                    return Err(ParseError::msg(
+                        "too many '[{' in filter directive, expected 0 or 1",
+                    ));
+                }
+
+                if !maybe_fields.ends_with("}]") {
+                    return Err(ParseError::msg("expected fields list to end with '}]'"));
+                }
+
+                let fields = maybe_fields
+                    .trim_end_matches("}]")
+                    .split(',')
+                    .filter_map(|s| {
+                        if s.is_empty() {
+                            None
+                        } else {
+                            Some(String::from(s))
+                        }
+                    });
+                field_names.extend(fields);
+            };
+            let level = part1.parse()?;
+            return Ok(Self {
+                level,
+                field_names,
+                target,
+            });
+        }
+
+        // Okay, the part after the `=` was empty, the directive is either a
+        // bare level or a bare target.
+        // * `foo`
+        // * `info`
+        Ok(match part0.parse::<LevelFilter>() {
+            Ok(level) => Self {
+                level,
+                target: None,
+                field_names: Vec::new(),
+            },
+            Err(_) => Self {
+                target: Some(String::from(part0)),
+                level: LevelFilter::TRACE,
+                field_names: Vec::new(),
+            },
+        })
+    }
+}
+
+// === impl ParseError ===
+
+impl ParseError {
+    #[cfg(all(feature = "std", feature = "env-filter"))]
+    pub(crate) fn new() -> Self {
+        ParseError {
+            kind: ParseErrorKind::Other(None),
+        }
+    }
+
+    pub(crate) fn msg(s: &'static str) -> Self {
+        ParseError {
+            kind: ParseErrorKind::Other(Some(s)),
+        }
+    }
+}
+
+impl fmt::Display for ParseError {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        match self.kind {
+            ParseErrorKind::Other(None) => f.pad("invalid filter directive"),
+            ParseErrorKind::Other(Some(msg)) => write!(f, "invalid filter directive: {}", msg),
+            ParseErrorKind::Level(ref l) => l.fmt(f),
+            #[cfg(feature = "std")]
+            ParseErrorKind::Field(ref e) => write!(f, "invalid field filter: {}", e),
+        }
+    }
+}
+
+#[cfg(feature = "std")]
+impl std::error::Error for ParseError {
+    fn description(&self) -> &str {
+        "invalid filter directive"
+    }
+
+    fn source(&self) -> Option<&(dyn std::error::Error + 'static)> {
+        match self.kind {
+            ParseErrorKind::Other(_) => None,
+            ParseErrorKind::Level(ref l) => Some(l),
+            ParseErrorKind::Field(ref n) => Some(n.as_ref()),
+        }
+    }
+}
+
+#[cfg(feature = "std")]
+impl From<Box<dyn std::error::Error + Send + Sync>> for ParseError {
+    fn from(e: Box<dyn std::error::Error + Send + Sync>) -> Self {
+        Self {
+            kind: ParseErrorKind::Field(e),
+        }
+    }
+}
+
+impl From<level::ParseError> for ParseError {
+    fn from(l: level::ParseError) -> Self {
+        Self {
+            kind: ParseErrorKind::Level(l),
+        }
+    }
+}

vendor/tracing-subscriber/src/filter/env/builder.rs

@@ -0,0 +1,351 @@
+use super::{
+    directive::{self, Directive},
+    EnvFilter, FromEnvError,
+};
+use crate::sync::RwLock;
+use std::env;
+use thread_local::ThreadLocal;
+use tracing::level_filters::STATIC_MAX_LEVEL;
+
+/// A [builder] for constructing new [`EnvFilter`]s.
+///
+/// [builder]: https://rust-unofficial.github.io/patterns/patterns/creational/builder.html
+#[derive(Debug, Clone)]
+#[must_use]
+pub struct Builder {
+    regex: bool,
+    env: Option<String>,
+    default_directive: Option<Directive>,
+}
+
+impl Builder {
+    /// Sets whether span field values can be matched with regular expressions.
+    ///
+    /// If this is `true`, field filter directives will be interpreted as
+    /// regular expressions if they are not able to be interpreted as a `bool`,
+    /// `i64`, `u64`, or `f64` literal. If this is `false,` those field values
+    /// will be interpreted as literal [`std::fmt::Debug`] output instead.
+    ///
+    /// By default, regular expressions are enabled.
+    ///
+    /// **Note**: when [`EnvFilter`]s are constructed from untrusted inputs,
+    /// disabling regular expressions is strongly encouraged.
+    pub fn with_regex(self, regex: bool) -> Self {
+        Self { regex, ..self }
+    }
+
+    /// Sets a default [filtering directive] that will be added to the filter if
+    /// the parsed string or environment variable contains no filter directives.
+    ///
+    /// By default, there is no default directive.
+    ///
+    /// # Examples
+    ///
+    /// If [`parse`], [`parse_lossy`], [`from_env`], or [`from_env_lossy`] are
+    /// called with an empty string or environment variable, the default
+    /// directive is used instead:
+    ///
+    /// ```rust
+    /// # fn main() -> Result<(), Box<dyn std::error::Error>> {
+    /// use tracing_subscriber::filter::{EnvFilter, LevelFilter};
+    ///
+    /// let filter = EnvFilter::builder()
+    ///     .with_default_directive(LevelFilter::INFO.into())
+    ///     .parse("")?;
+    ///
+    /// assert_eq!(format!("{}", filter), "info");
+    /// # Ok(()) }
+    /// ```
+    ///
+    /// Note that the `lossy` variants ([`parse_lossy`] and [`from_env_lossy`])
+    /// will ignore any invalid directives. If all directives in a filter
+    /// string or environment variable are invalid, those methods will also use
+    /// the default directive:
+    ///
+    /// ```rust
+    /// use tracing_subscriber::filter::{EnvFilter, LevelFilter};
+    ///
+    /// let filter = EnvFilter::builder()
+    ///     .with_default_directive(LevelFilter::INFO.into())
+    ///     .parse_lossy("some_target=fake level,foo::bar=lolwut");
+    ///
+    /// assert_eq!(format!("{}", filter), "info");
+    /// ```
+    ///
+    ///
+    /// If the string or environment variable contains valid filtering
+    /// directives, the default directive is not used:
+    ///
+    /// ```rust
+    /// use tracing_subscriber::filter::{EnvFilter, LevelFilter};
+    ///
+    /// let filter = EnvFilter::builder()
+    ///     .with_default_directive(LevelFilter::INFO.into())
+    ///     .parse_lossy("foo=trace");
+    ///
+    /// // The default directive is *not* used:
+    /// assert_eq!(format!("{}", filter), "foo=trace");
+    /// ```
+    ///
+    /// Parsing a more complex default directive from a string:
+    ///
+    /// ```rust
+    /// # fn main() -> Result<(), Box<dyn std::error::Error>> {
+    /// use tracing_subscriber::filter::{EnvFilter, LevelFilter};
+    ///
+    /// let default = "myapp=debug".parse()
+    ///     .expect("hard-coded default directive should be valid");
+    ///
+    /// let filter = EnvFilter::builder()
+    ///     .with_default_directive(default)
+    ///     .parse("")?;
+    ///
+    /// assert_eq!(format!("{}", filter), "myapp=debug");
+    /// # Ok(()) }
+    /// ```
+    ///
+    /// [`parse_lossy`]: Self::parse_lossy
+    /// [`from_env_lossy`]: Self::from_env_lossy
+    /// [`parse`]: Self::parse
+    /// [`from_env`]: Self::from_env
+    pub fn with_default_directive(self, default_directive: Directive) -> Self {
+        Self {
+            default_directive: Some(default_directive),
+            ..self
+        }
+    }
+
+    /// Sets the name of the environment variable used by the [`from_env`],
+    /// [`from_env_lossy`], and [`try_from_env`] methods.
+    ///
+    /// By default, this is the value of [`EnvFilter::DEFAULT_ENV`]
+    /// (`RUST_LOG`).
+    ///
+    /// [`from_env`]: Self::from_env
+    /// [`from_env_lossy`]: Self::from_env_lossy
+    /// [`try_from_env`]: Self::try_from_env
+    pub fn with_env_var(self, var: impl ToString) -> Self {
+        Self {
+            env: Some(var.to_string()),
+            ..self
+        }
+    }
+
+    /// Returns a new [`EnvFilter`] from the directives in the given string,
+    /// *ignoring* any that are invalid.
+    ///
+    /// If `parse_lossy` is called with an empty string, then the
+    /// [default directive] is used instead.
+    ///
+    /// [default directive]: Self::with_default_directive
+    pub fn parse_lossy<S: AsRef<str>>(&self, dirs: S) -> EnvFilter {
+        let directives = dirs
+            .as_ref()
+            .split(',')
+            .filter(|s| !s.is_empty())
+            .filter_map(|s| match Directive::parse(s, self.regex) {
+                Ok(d) => Some(d),
+                Err(err) => {
+                    eprintln!("ignoring `{}`: {}", s, err);
+                    None
+                }
+            });
+        self.from_directives(directives)
+    }
+
+    /// Returns a new [`EnvFilter`] from the directives in the given string,
+    /// or an error if any are invalid.
+    ///
+    /// If `parse` is called with an empty string, then the [default directive]
+    /// is used instead.
+    ///
+    /// [default directive]: Self::with_default_directive
+    pub fn parse<S: AsRef<str>>(&self, dirs: S) -> Result<EnvFilter, directive::ParseError> {
+        let dirs = dirs.as_ref();
+        if dirs.is_empty() {
+            return Ok(self.from_directives(std::iter::empty()));
+        }
+        let directives = dirs
+            .split(',')
+            .filter(|s| !s.is_empty())
+            .map(|s| Directive::parse(s, self.regex))
+            .collect::<Result<Vec<_>, _>>()?;
+        Ok(self.from_directives(directives))
+    }
+
+    /// Returns a new [`EnvFilter`] from the directives in the configured
+    /// environment variable, ignoring any directives that are invalid.
+    ///
+    /// If the environment variable is empty, then the [default directive]
+    /// is used instead.
+    ///
+    /// [default directive]: Self::with_default_directive
+    pub fn from_env_lossy(&self) -> EnvFilter {
+        let var = env::var(self.env_var_name()).unwrap_or_default();
+        self.parse_lossy(var)
+    }
+
+    /// Returns a new [`EnvFilter`] from the directives in the configured
+    /// environment variable. If the environment variable is unset, no directive is added.
+    ///
+    /// An error is returned if the environment contains invalid directives.
+    ///
+    /// If the environment variable is empty, then the [default directive]
+    /// is used instead.
+    ///
+    /// [default directive]: Self::with_default_directive
+    pub fn from_env(&self) -> Result<EnvFilter, FromEnvError> {
+        let var = env::var(self.env_var_name()).unwrap_or_default();
+        self.parse(var).map_err(Into::into)
+    }
+
+    /// Returns a new [`EnvFilter`] from the directives in the configured
+    /// environment variable, or an error if the environment variable is not set
+    /// or contains invalid directives.
+    ///
+    /// If the environment variable is empty, then the [default directive]
+    /// is used instead.
+    ///
+    /// [default directive]: Self::with_default_directive
+    pub fn try_from_env(&self) -> Result<EnvFilter, FromEnvError> {
+        let var = env::var(self.env_var_name())?;
+        self.parse(var).map_err(Into::into)
+    }
+
+    // TODO(eliza): consider making this a public API?
+    // Clippy doesn't love this naming, because it suggests that `from_` methods
+    // should not take a `Self`...but in this case, it's the `EnvFilter` that is
+    // being constructed "from" the directives, rather than the builder itself.
+    #[allow(clippy::wrong_self_convention)]
+    pub(super) fn from_directives(
+        &self,
+        directives: impl IntoIterator<Item = Directive>,
+    ) -> EnvFilter {
+        use tracing::Level;
+
+        let mut directives: Vec<_> = directives.into_iter().collect();
+        let mut disabled = Vec::new();
+        for directive in &mut directives {
+            if directive.level > STATIC_MAX_LEVEL {
+                disabled.push(directive.clone());
+            }
+            if !self.regex {
+                directive.deregexify();
+            }
+        }
+
+        if !disabled.is_empty() {
+            #[cfg(feature = "nu-ansi-term")]
+            use nu_ansi_term::{Color, Style};
+            // NOTE: We can't use a configured `MakeWriter` because the EnvFilter
+            // has no knowledge of any underlying subscriber or subscriber, which
+            // may or may not use a `MakeWriter`.
+            let warn = |msg: &str| {
+                #[cfg(not(feature = "nu-ansi-term"))]
+                let msg = format!("warning: {}", msg);
+                #[cfg(feature = "nu-ansi-term")]
+                let msg = {
+                    let bold = Style::new().bold();
+                    let mut warning = Color::Yellow.paint("warning");
+                    warning.style_ref_mut().is_bold = true;
+                    format!("{}{} {}", warning, bold.paint(":"), bold.paint(msg))
+                };
+                eprintln!("{}", msg);
+            };
+            let ctx_prefixed = |prefix: &str, msg: &str| {
+                #[cfg(not(feature = "nu-ansi-term"))]
+                let msg = format!("{} {}", prefix, msg);
+                #[cfg(feature = "nu-ansi-term")]
+                let msg = {
+                    let mut equal = Color::Fixed(21).paint("="); // dark blue
+                    equal.style_ref_mut().is_bold = true;
+                    format!(" {} {} {}", equal, Style::new().bold().paint(prefix), msg)
+                };
+                eprintln!("{}", msg);
+            };
+            let ctx_help = |msg| ctx_prefixed("help:", msg);
+            let ctx_note = |msg| ctx_prefixed("note:", msg);
+            let ctx = |msg: &str| {
+                #[cfg(not(feature = "nu-ansi-term"))]
+                let msg = format!("note: {}", msg);
+                #[cfg(feature = "nu-ansi-term")]
+                let msg = {
+                    let mut pipe = Color::Fixed(21).paint("|");
+                    pipe.style_ref_mut().is_bold = true;
+                    format!(" {} {}", pipe, msg)
+                };
+                eprintln!("{}", msg);
+            };
+            warn("some trace filter directives would enable traces that are disabled statically");
+            for directive in disabled {
+                let target = if let Some(target) = &directive.target {
+                    format!("the `{}` target", target)
+                } else {
+                    "all targets".into()
+                };
+                let level = directive
+                    .level
+                    .into_level()
+                    .expect("=off would not have enabled any filters");
+                ctx(&format!(
+                    "`{}` would enable the {} level for {}",
+                    directive, level, target
+                ));
+            }
+            ctx_note(&format!("the static max level is `{}`", STATIC_MAX_LEVEL));
+            let help_msg = || {
+                let (feature, filter) = match STATIC_MAX_LEVEL.into_level() {
+                    Some(Level::TRACE) => unreachable!(
+                        "if the max level is trace, no static filtering features are enabled"
+                    ),
+                    Some(Level::DEBUG) => ("max_level_debug", Level::TRACE),
+                    Some(Level::INFO) => ("max_level_info", Level::DEBUG),
+                    Some(Level::WARN) => ("max_level_warn", Level::INFO),
+                    Some(Level::ERROR) => ("max_level_error", Level::WARN),
+                    None => return ("max_level_off", String::new()),
+                };
+                (feature, format!("{} ", filter))
+            };
+            let (feature, earlier_level) = help_msg();
+            ctx_help(&format!(
+                "to enable {}logging, remove the `{}` feature from the `tracing` crate",
+                earlier_level, feature
+            ));
+        }
+
+        let (dynamics, statics) = Directive::make_tables(directives);
+        let has_dynamics = !dynamics.is_empty();
+
+        let mut filter = EnvFilter {
+            statics,
+            dynamics,
+            has_dynamics,
+            by_id: RwLock::new(Default::default()),
+            by_cs: RwLock::new(Default::default()),
+            scope: ThreadLocal::new(),
+            regex: self.regex,
+        };
+
+        if !has_dynamics && filter.statics.is_empty() {
+            if let Some(ref default) = self.default_directive {
+                filter = filter.add_directive(default.clone());
+            }
+        }
+
+        filter
+    }
+
+    fn env_var_name(&self) -> &str {
+        self.env.as_deref().unwrap_or(EnvFilter::DEFAULT_ENV)
+    }
+}
+
+impl Default for Builder {
+    fn default() -> Self {
+        Self {
+            regex: true,
+            env: None,
+            default_directive: None,
+        }
+    }
+}

vendor/tracing-subscriber/src/filter/env/directive.rs

@@ -0,0 +1,888 @@
+pub(crate) use crate::filter::directive::{FilterVec, ParseError, StaticDirective};
+use crate::filter::{
+    directive::{DirectiveSet, Match},
+    env::{field, FieldMap},
+    level::LevelFilter,
+};
+use std::{cmp::Ordering, fmt, iter::FromIterator, str::FromStr};
+use tracing_core::{span, Level, Metadata};
+
+/// A single filtering directive.
+// TODO(eliza): add a builder for programmatically constructing directives?
+#[derive(Clone, Debug, Eq, PartialEq)]
+#[cfg_attr(docsrs, doc(cfg(feature = "env-filter")))]
+pub struct Directive {
+    in_span: Option<String>,
+    fields: Vec<field::Match>,
+    pub(crate) target: Option<String>,
+    pub(crate) level: LevelFilter,
+}
+
+/// A set of dynamic filtering directives.
+pub(super) type Dynamics = DirectiveSet<Directive>;
+
+/// A set of static filtering directives.
+pub(super) type Statics = DirectiveSet<StaticDirective>;
+
+pub(crate) type CallsiteMatcher = MatchSet<field::CallsiteMatch>;
+pub(crate) type SpanMatcher = MatchSet<field::SpanMatch>;
+
+#[derive(Debug, PartialEq, Eq)]
+pub(crate) struct MatchSet<T> {
+    field_matches: FilterVec<T>,
+    base_level: LevelFilter,
+}
+
+impl Directive {
+    pub(super) fn has_name(&self) -> bool {
+        self.in_span.is_some()
+    }
+
+    pub(super) fn has_fields(&self) -> bool {
+        !self.fields.is_empty()
+    }
+
+    pub(super) fn to_static(&self) -> Option<StaticDirective> {
+        if !self.is_static() {
+            return None;
+        }
+
+        // TODO(eliza): these strings are all immutable; we should consider
+        // `Arc`ing them to make this more efficient...
+        let field_names = self.fields.iter().map(field::Match::name).collect();
+
+        Some(StaticDirective::new(
+            self.target.clone(),
+            field_names,
+            self.level,
+        ))
+    }
+
+    fn is_static(&self) -> bool {
+        !self.has_name() && !self.fields.iter().any(field::Match::has_value)
+    }
+
+    pub(super) fn is_dynamic(&self) -> bool {
+        self.has_name() || self.has_fields()
+    }
+
+    pub(crate) fn field_matcher(&self, meta: &Metadata<'_>) -> Option<field::CallsiteMatch> {
+        let fieldset = meta.fields();
+        let fields = self
+            .fields
+            .iter()
+            .filter_map(
+                |field::Match {
+                     ref name,
+                     ref value,
+                 }| {
+                    if let Some(field) = fieldset.field(name) {
+                        let value = value.as_ref().cloned()?;
+                        Some(Ok((field, value)))
+                    } else {
+                        Some(Err(()))
+                    }
+                },
+            )
+            .collect::<Result<FieldMap<_>, ()>>()
+            .ok()?;
+        Some(field::CallsiteMatch {
+            fields,
+            level: self.level,
+        })
+    }
+
+    pub(super) fn make_tables(
+        directives: impl IntoIterator<Item = Directive>,
+    ) -> (Dynamics, Statics) {
+        // TODO(eliza): this could be made more efficient...
+        let (dyns, stats): (Vec<Directive>, Vec<Directive>) =
+            directives.into_iter().partition(Directive::is_dynamic);
+        let statics = stats
+            .into_iter()
+            .filter_map(|d| d.to_static())
+            .chain(dyns.iter().filter_map(Directive::to_static))
+            .collect();
+        (Dynamics::from_iter(dyns), statics)
+    }
+
+    pub(super) fn deregexify(&mut self) {
+        for field in &mut self.fields {
+            field.value = match field.value.take() {
+                Some(field::ValueMatch::Pat(pat)) => {
+                    Some(field::ValueMatch::Debug(pat.into_debug_match()))
+                }
+                x => x,
+            }
+        }
+    }
+
+    pub(super) fn parse(from: &str, regex: bool) -> Result<Self, ParseError> {
+        let mut cur = Self {
+            level: LevelFilter::TRACE,
+            target: None,
+            in_span: None,
+            fields: Vec::new(),
+        };
+
+        #[derive(Debug)]
+        enum ParseState {
+            Start,
+            LevelOrTarget { start: usize },
+            Span { span_start: usize },
+            Field { field_start: usize },
+            Fields,
+            Target,
+            Level { level_start: usize },
+            Complete,
+        }
+
+        use ParseState::*;
+        let mut state = Start;
+        for (i, c) in from.trim().char_indices() {
+            state = match (state, c) {
+                (Start, '[') => Span { span_start: i + 1 },
+                (Start, c) if !['-', ':', '_'].contains(&c) && !c.is_alphanumeric() => {
+                    return Err(ParseError::new())
+                }
+                (Start, _) => LevelOrTarget { start: i },
+                (LevelOrTarget { start }, '=') => {
+                    cur.target = Some(from[start..i].to_owned());
+                    Level { level_start: i + 1 }
+                }
+                (LevelOrTarget { start }, '[') => {
+                    cur.target = Some(from[start..i].to_owned());
+                    Span { span_start: i + 1 }
+                }
+                (LevelOrTarget { start }, ',') => {
+                    let (level, target) = match &from[start..] {
+                        "" => (LevelFilter::TRACE, None),
+                        level_or_target => match LevelFilter::from_str(level_or_target) {
+                            Ok(level) => (level, None),
+                            Err(_) => (LevelFilter::TRACE, Some(level_or_target.to_owned())),
+                        },
+                    };
+
+                    cur.level = level;
+                    cur.target = target;
+                    Complete
+                }
+                (state @ LevelOrTarget { .. }, _) => state,
+                (Target, '=') => Level { level_start: i + 1 },
+                (Span { span_start }, ']') => {
+                    cur.in_span = Some(from[span_start..i].to_owned());
+                    Target
+                }
+                (Span { span_start }, '{') => {
+                    cur.in_span = match &from[span_start..i] {
+                        "" => None,
+                        _ => Some(from[span_start..i].to_owned()),
+                    };
+                    Field { field_start: i + 1 }
+                }
+                (state @ Span { .. }, _) => state,
+                (Field { field_start }, '}') => {
+                    cur.fields.push(match &from[field_start..i] {
+                        "" => return Err(ParseError::new()),
+                        field => field::Match::parse(field, regex)?,
+                    });
+                    Fields
+                }
+                (Field { field_start }, ',') => {
+                    cur.fields.push(match &from[field_start..i] {
+                        "" => return Err(ParseError::new()),
+                        field => field::Match::parse(field, regex)?,
+                    });
+                    Field { field_start: i + 1 }
+                }
+                (state @ Field { .. }, _) => state,
+                (Fields, ']') => Target,
+                (Level { level_start }, ',') => {
+                    cur.level = match &from[level_start..i] {
+                        "" => LevelFilter::TRACE,
+                        level => LevelFilter::from_str(level)?,
+                    };
+                    Complete
+                }
+                (state @ Level { .. }, _) => state,
+                _ => return Err(ParseError::new()),
+            };
+        }
+
+        match state {
+            LevelOrTarget { start } => {
+                let (level, target) = match &from[start..] {
+                    "" => (LevelFilter::TRACE, None),
+                    level_or_target => match LevelFilter::from_str(level_or_target) {
+                        Ok(level) => (level, None),
+                        // Setting the target without the level enables every level for that target
+                        Err(_) => (LevelFilter::TRACE, Some(level_or_target.to_owned())),
+                    },
+                };
+
+                cur.level = level;
+                cur.target = target;
+            }
+            Level { level_start } => {
+                cur.level = match &from[level_start..] {
+                    "" => LevelFilter::TRACE,
+                    level => LevelFilter::from_str(level)?,
+                };
+            }
+            Target | Complete => {}
+            _ => return Err(ParseError::new()),
+        };
+
+        Ok(cur)
+    }
+}
+
+impl Match for Directive {
+    fn cares_about(&self, meta: &Metadata<'_>) -> bool {
+        // Does this directive have a target filter, and does it match the
+        // metadata's target?
+        if let Some(ref target) = self.target {
+            if !meta.target().starts_with(&target[..]) {
+                return false;
+            }
+        }
+
+        // Do we have a name filter, and does it match the metadata's name?
+        // TODO(eliza): put name globbing here?
+        if let Some(ref name) = self.in_span {
+            if name != meta.name() {
+                return false;
+            }
+        }
+
+        // Does the metadata define all the fields that this directive cares about?
+        let actual_fields = meta.fields();
+        for expected_field in &self.fields {
+            // Does the actual field set (from the metadata) contain this field?
+            if actual_fields.field(&expected_field.name).is_none() {
+                return false;
+            }
+        }
+
+        true
+    }
+
+    fn level(&self) -> &LevelFilter {
+        &self.level
+    }
+}
+
+impl FromStr for Directive {
+    type Err = ParseError;
+    fn from_str(from: &str) -> Result<Self, Self::Err> {
+        Directive::parse(from, true)
+    }
+}
+
+impl Default for Directive {
+    fn default() -> Self {
+        Directive {
+            level: LevelFilter::OFF,
+            target: None,
+            in_span: None,
+            fields: Vec::new(),
+        }
+    }
+}
+
+impl PartialOrd for Directive {
+    fn partial_cmp(&self, other: &Directive) -> Option<Ordering> {
+        Some(self.cmp(other))
+    }
+}
+
+impl Ord for Directive {
+    fn cmp(&self, other: &Directive) -> Ordering {
+        // We attempt to order directives by how "specific" they are. This
+        // ensures that we try the most specific directives first when
+        // attempting to match a piece of metadata.
+
+        // First, we compare based on whether a target is specified, and the
+        // lengths of those targets if both have targets.
+        let ordering = self
+            .target
+            .as_ref()
+            .map(String::len)
+            .cmp(&other.target.as_ref().map(String::len))
+            // Next compare based on the presence of span names.
+            .then_with(|| self.in_span.is_some().cmp(&other.in_span.is_some()))
+            // Then we compare how many fields are defined by each
+            // directive.
+            .then_with(|| self.fields.len().cmp(&other.fields.len()))
+            // Finally, we fall back to lexicographical ordering if the directives are
+            // equally specific. Although this is no longer semantically important,
+            // we need to define a total ordering to determine the directive's place
+            // in the BTreeMap.
+            .then_with(|| {
+                self.target
+                    .cmp(&other.target)
+                    .then_with(|| self.in_span.cmp(&other.in_span))
+                    .then_with(|| self.fields[..].cmp(&other.fields[..]))
+            })
+            .reverse();
+
+        #[cfg(debug_assertions)]
+        {
+            if ordering == Ordering::Equal {
+                debug_assert_eq!(
+                    self.target, other.target,
+                    "invariant violated: Ordering::Equal must imply a.target == b.target"
+                );
+                debug_assert_eq!(
+                    self.in_span, other.in_span,
+                    "invariant violated: Ordering::Equal must imply a.in_span == b.in_span"
+                );
+                debug_assert_eq!(
+                    self.fields, other.fields,
+                    "invariant violated: Ordering::Equal must imply a.fields == b.fields"
+                );
+            }
+        }
+
+        ordering
+    }
+}
+
+impl fmt::Display for Directive {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        let mut wrote_any = false;
+        if let Some(ref target) = self.target {
+            fmt::Display::fmt(target, f)?;
+            wrote_any = true;
+        }
+
+        if self.in_span.is_some() || !self.fields.is_empty() {
+            f.write_str("[")?;
+
+            if let Some(ref span) = self.in_span {
+                fmt::Display::fmt(span, f)?;
+            }
+
+            let mut fields = self.fields.iter();
+            if let Some(field) = fields.next() {
+                write!(f, "{{{}", field)?;
+                for field in fields {
+                    write!(f, ",{}", field)?;
+                }
+                f.write_str("}")?;
+            }
+
+            f.write_str("]")?;
+            wrote_any = true;
+        }
+
+        if wrote_any {
+            f.write_str("=")?;
+        }
+
+        fmt::Display::fmt(&self.level, f)
+    }
+}
+
+impl From<LevelFilter> for Directive {
+    fn from(level: LevelFilter) -> Self {
+        Self {
+            level,
+            ..Self::default()
+        }
+    }
+}
+
+impl From<Level> for Directive {
+    fn from(level: Level) -> Self {
+        LevelFilter::from_level(level).into()
+    }
+}
+
+// === impl Dynamics ===
+
+impl Dynamics {
+    pub(crate) fn matcher(&self, metadata: &Metadata<'_>) -> Option<CallsiteMatcher> {
+        let mut base_level = None;
+        let field_matches = self
+            .directives_for(metadata)
+            .filter_map(|d| {
+                if let Some(f) = d.field_matcher(metadata) {
+                    return Some(f);
+                }
+                match base_level {
+                    Some(ref b) if d.level > *b => base_level = Some(d.level),
+                    None => base_level = Some(d.level),
+                    _ => {}
+                }
+                None
+            })
+            .collect();
+
+        if let Some(base_level) = base_level {
+            Some(CallsiteMatcher {
+                field_matches,
+                base_level,
+            })
+        } else if !field_matches.is_empty() {
+            Some(CallsiteMatcher {
+                field_matches,
+                base_level: base_level.unwrap_or(LevelFilter::OFF),
+            })
+        } else {
+            None
+        }
+    }
+
+    pub(crate) fn has_value_filters(&self) -> bool {
+        self.directives()
+            .any(|d| d.fields.iter().any(|f| f.value.is_some()))
+    }
+}
+
+// ===== impl DynamicMatch =====
+
+impl CallsiteMatcher {
+    /// Create a new `SpanMatch` for a given instance of the matched callsite.
+    pub(crate) fn to_span_match(&self, attrs: &span::Attributes<'_>) -> SpanMatcher {
+        let field_matches = self
+            .field_matches
+            .iter()
+            .map(|m| {
+                let m = m.to_span_match();
+                attrs.record(&mut m.visitor());
+                m
+            })
+            .collect();
+        SpanMatcher {
+            field_matches,
+            base_level: self.base_level,
+        }
+    }
+}
+
+impl SpanMatcher {
+    /// Returns the level currently enabled for this callsite.
+    pub(crate) fn level(&self) -> LevelFilter {
+        self.field_matches
+            .iter()
+            .filter_map(field::SpanMatch::filter)
+            .max()
+            .unwrap_or(self.base_level)
+    }
+
+    pub(crate) fn record_update(&self, record: &span::Record<'_>) {
+        for m in &self.field_matches {
+            record.record(&mut m.visitor())
+        }
+    }
+}
+
+#[cfg(test)]
+mod test {
+    use super::*;
+
+    fn parse_directives(dirs: impl AsRef<str>) -> Vec<Directive> {
+        dirs.as_ref()
+            .split(',')
+            .filter_map(|s| s.parse().ok())
+            .collect()
+    }
+
+    fn expect_parse(dirs: impl AsRef<str>) -> Vec<Directive> {
+        dirs.as_ref()
+            .split(',')
+            .map(|s| {
+                s.parse()
+                    .unwrap_or_else(|err| panic!("directive '{:?}' should parse: {}", s, err))
+            })
+            .collect()
+    }
+
+    #[test]
+    fn directive_ordering_by_target_len() {
+        // TODO(eliza): it would be nice to have a property-based test for this
+        // instead.
+        let mut dirs = expect_parse(
+            "foo::bar=debug,foo::bar::baz=trace,foo=info,a_really_long_name_with_no_colons=warn",
+        );
+        dirs.sort_unstable();
+
+        let expected = vec![
+            "a_really_long_name_with_no_colons",
+            "foo::bar::baz",
+            "foo::bar",
+            "foo",
+        ];
+        let sorted = dirs
+            .iter()
+            .map(|d| d.target.as_ref().unwrap())
+            .collect::<Vec<_>>();
+
+        assert_eq!(expected, sorted);
+    }
+    #[test]
+    fn directive_ordering_by_span() {
+        // TODO(eliza): it would be nice to have a property-based test for this
+        // instead.
+        let mut dirs = expect_parse("bar[span]=trace,foo=debug,baz::quux=info,a[span]=warn");
+        dirs.sort_unstable();
+
+        let expected = vec!["baz::quux", "bar", "foo", "a"];
+        let sorted = dirs
+            .iter()
+            .map(|d| d.target.as_ref().unwrap())
+            .collect::<Vec<_>>();
+
+        assert_eq!(expected, sorted);
+    }
+
+    #[test]
+    fn directive_ordering_uses_lexicographic_when_equal() {
+        // TODO(eliza): it would be nice to have a property-based test for this
+        // instead.
+        let mut dirs = expect_parse("span[b]=debug,b=debug,a=trace,c=info,span[a]=info");
+        dirs.sort_unstable();
+
+        let expected = vec![
+            ("span", Some("b")),
+            ("span", Some("a")),
+            ("c", None),
+            ("b", None),
+            ("a", None),
+        ];
+        let sorted = dirs
+            .iter()
+            .map(|d| {
+                (
+                    d.target.as_ref().unwrap().as_ref(),
+                    d.in_span.as_ref().map(String::as_ref),
+                )
+            })
+            .collect::<Vec<_>>();
+
+        assert_eq!(expected, sorted);
+    }
+
+    // TODO: this test requires the parser to support directives with multiple
+    // fields, which it currently can't handle. We should enable this test when
+    // that's implemented.
+    #[test]
+    #[ignore]
+    fn directive_ordering_by_field_num() {
+        // TODO(eliza): it would be nice to have a property-based test for this
+        // instead.
+        let mut dirs = expect_parse(
+            "b[{foo,bar}]=info,c[{baz,quuux,quuux}]=debug,a[{foo}]=warn,bar[{field}]=trace,foo=debug,baz::quux=info"
+        );
+        dirs.sort_unstable();
+
+        let expected = vec!["baz::quux", "bar", "foo", "c", "b", "a"];
+        let sorted = dirs
+            .iter()
+            .map(|d| d.target.as_ref().unwrap())
+            .collect::<Vec<_>>();
+
+        assert_eq!(expected, sorted);
+    }
+
+    #[test]
+    fn parse_directives_ralith() {
+        let dirs = parse_directives("common=trace,server=trace");
+        assert_eq!(dirs.len(), 2, "\nparsed: {:#?}", dirs);
+        assert_eq!(dirs[0].target, Some("common".to_string()));
+        assert_eq!(dirs[0].level, LevelFilter::TRACE);
+        assert_eq!(dirs[0].in_span, None);
+
+        assert_eq!(dirs[1].target, Some("server".to_string()));
+        assert_eq!(dirs[1].level, LevelFilter::TRACE);
+        assert_eq!(dirs[1].in_span, None);
+    }
+
+    #[test]
+    fn parse_directives_ralith_uc() {
+        let dirs = parse_directives("common=INFO,server=DEBUG");
+        assert_eq!(dirs.len(), 2, "\nparsed: {:#?}", dirs);
+        assert_eq!(dirs[0].target, Some("common".to_string()));
+        assert_eq!(dirs[0].level, LevelFilter::INFO);
+        assert_eq!(dirs[0].in_span, None);
+
+        assert_eq!(dirs[1].target, Some("server".to_string()));
+        assert_eq!(dirs[1].level, LevelFilter::DEBUG);
+        assert_eq!(dirs[1].in_span, None);
+    }
+
+    #[test]
+    fn parse_directives_ralith_mixed() {
+        let dirs = parse_directives("common=iNfo,server=dEbUg");
+        assert_eq!(dirs.len(), 2, "\nparsed: {:#?}", dirs);
+        assert_eq!(dirs[0].target, Some("common".to_string()));
+        assert_eq!(dirs[0].level, LevelFilter::INFO);
+        assert_eq!(dirs[0].in_span, None);
+
+        assert_eq!(dirs[1].target, Some("server".to_string()));
+        assert_eq!(dirs[1].level, LevelFilter::DEBUG);
+        assert_eq!(dirs[1].in_span, None);
+    }
+
+    #[test]
+    fn parse_directives_valid() {
+        let dirs = parse_directives("crate1::mod1=error,crate1::mod2,crate2=debug,crate3=off");
+        assert_eq!(dirs.len(), 4, "\nparsed: {:#?}", dirs);
+        assert_eq!(dirs[0].target, Some("crate1::mod1".to_string()));
+        assert_eq!(dirs[0].level, LevelFilter::ERROR);
+        assert_eq!(dirs[0].in_span, None);
+
+        assert_eq!(dirs[1].target, Some("crate1::mod2".to_string()));
+        assert_eq!(dirs[1].level, LevelFilter::TRACE);
+        assert_eq!(dirs[1].in_span, None);
+
+        assert_eq!(dirs[2].target, Some("crate2".to_string()));
+        assert_eq!(dirs[2].level, LevelFilter::DEBUG);
+        assert_eq!(dirs[2].in_span, None);
+
+        assert_eq!(dirs[3].target, Some("crate3".to_string()));
+        assert_eq!(dirs[3].level, LevelFilter::OFF);
+        assert_eq!(dirs[3].in_span, None);
+    }
+
+    #[test]
+
+    fn parse_level_directives() {
+        let dirs = parse_directives(
+            "crate1::mod1=error,crate1::mod2=warn,crate1::mod2::mod3=info,\
+             crate2=debug,crate3=trace,crate3::mod2::mod1=off",
+        );
+        assert_eq!(dirs.len(), 6, "\nparsed: {:#?}", dirs);
+        assert_eq!(dirs[0].target, Some("crate1::mod1".to_string()));
+        assert_eq!(dirs[0].level, LevelFilter::ERROR);
+        assert_eq!(dirs[0].in_span, None);
+
+        assert_eq!(dirs[1].target, Some("crate1::mod2".to_string()));
+        assert_eq!(dirs[1].level, LevelFilter::WARN);
+        assert_eq!(dirs[1].in_span, None);
+
+        assert_eq!(dirs[2].target, Some("crate1::mod2::mod3".to_string()));
+        assert_eq!(dirs[2].level, LevelFilter::INFO);
+        assert_eq!(dirs[2].in_span, None);
+
+        assert_eq!(dirs[3].target, Some("crate2".to_string()));
+        assert_eq!(dirs[3].level, LevelFilter::DEBUG);
+        assert_eq!(dirs[3].in_span, None);
+
+        assert_eq!(dirs[4].target, Some("crate3".to_string()));
+        assert_eq!(dirs[4].level, LevelFilter::TRACE);
+        assert_eq!(dirs[4].in_span, None);
+
+        assert_eq!(dirs[5].target, Some("crate3::mod2::mod1".to_string()));
+        assert_eq!(dirs[5].level, LevelFilter::OFF);
+        assert_eq!(dirs[5].in_span, None);
+    }
+
+    #[test]
+    fn parse_uppercase_level_directives() {
+        let dirs = parse_directives(
+            "crate1::mod1=ERROR,crate1::mod2=WARN,crate1::mod2::mod3=INFO,\
+             crate2=DEBUG,crate3=TRACE,crate3::mod2::mod1=OFF",
+        );
+        assert_eq!(dirs.len(), 6, "\nparsed: {:#?}", dirs);
+        assert_eq!(dirs[0].target, Some("crate1::mod1".to_string()));
+        assert_eq!(dirs[0].level, LevelFilter::ERROR);
+        assert_eq!(dirs[0].in_span, None);
+
+        assert_eq!(dirs[1].target, Some("crate1::mod2".to_string()));
+        assert_eq!(dirs[1].level, LevelFilter::WARN);
+        assert_eq!(dirs[1].in_span, None);
+
+        assert_eq!(dirs[2].target, Some("crate1::mod2::mod3".to_string()));
+        assert_eq!(dirs[2].level, LevelFilter::INFO);
+        assert_eq!(dirs[2].in_span, None);
+
+        assert_eq!(dirs[3].target, Some("crate2".to_string()));
+        assert_eq!(dirs[3].level, LevelFilter::DEBUG);
+        assert_eq!(dirs[3].in_span, None);
+
+        assert_eq!(dirs[4].target, Some("crate3".to_string()));
+        assert_eq!(dirs[4].level, LevelFilter::TRACE);
+        assert_eq!(dirs[4].in_span, None);
+
+        assert_eq!(dirs[5].target, Some("crate3::mod2::mod1".to_string()));
+        assert_eq!(dirs[5].level, LevelFilter::OFF);
+        assert_eq!(dirs[5].in_span, None);
+    }
+
+    #[test]
+    fn parse_numeric_level_directives() {
+        let dirs = parse_directives(
+            "crate1::mod1=1,crate1::mod2=2,crate1::mod2::mod3=3,crate2=4,\
+             crate3=5,crate3::mod2::mod1=0",
+        );
+        assert_eq!(dirs.len(), 6, "\nparsed: {:#?}", dirs);
+        assert_eq!(dirs[0].target, Some("crate1::mod1".to_string()));
+        assert_eq!(dirs[0].level, LevelFilter::ERROR);
+        assert_eq!(dirs[0].in_span, None);
+
+        assert_eq!(dirs[1].target, Some("crate1::mod2".to_string()));
+        assert_eq!(dirs[1].level, LevelFilter::WARN);
+        assert_eq!(dirs[1].in_span, None);
+
+        assert_eq!(dirs[2].target, Some("crate1::mod2::mod3".to_string()));
+        assert_eq!(dirs[2].level, LevelFilter::INFO);
+        assert_eq!(dirs[2].in_span, None);
+
+        assert_eq!(dirs[3].target, Some("crate2".to_string()));
+        assert_eq!(dirs[3].level, LevelFilter::DEBUG);
+        assert_eq!(dirs[3].in_span, None);
+
+        assert_eq!(dirs[4].target, Some("crate3".to_string()));
+        assert_eq!(dirs[4].level, LevelFilter::TRACE);
+        assert_eq!(dirs[4].in_span, None);
+
+        assert_eq!(dirs[5].target, Some("crate3::mod2::mod1".to_string()));
+        assert_eq!(dirs[5].level, LevelFilter::OFF);
+        assert_eq!(dirs[5].in_span, None);
+    }
+
+    #[test]
+    fn parse_directives_invalid_crate() {
+        // test parse_directives with multiple = in specification
+        let dirs = parse_directives("crate1::mod1=warn=info,crate2=debug");
+        assert_eq!(dirs.len(), 1, "\nparsed: {:#?}", dirs);
+        assert_eq!(dirs[0].target, Some("crate2".to_string()));
+        assert_eq!(dirs[0].level, LevelFilter::DEBUG);
+        assert_eq!(dirs[0].in_span, None);
+    }
+
+    #[test]
+    fn parse_directives_invalid_level() {
+        // test parse_directives with 'noNumber' as log level
+        let dirs = parse_directives("crate1::mod1=noNumber,crate2=debug");
+        assert_eq!(dirs.len(), 1, "\nparsed: {:#?}", dirs);
+        assert_eq!(dirs[0].target, Some("crate2".to_string()));
+        assert_eq!(dirs[0].level, LevelFilter::DEBUG);
+        assert_eq!(dirs[0].in_span, None);
+    }
+
+    #[test]
+    fn parse_directives_string_level() {
+        // test parse_directives with 'warn' as log level
+        let dirs = parse_directives("crate1::mod1=wrong,crate2=warn");
+        assert_eq!(dirs.len(), 1, "\nparsed: {:#?}", dirs);
+        assert_eq!(dirs[0].target, Some("crate2".to_string()));
+        assert_eq!(dirs[0].level, LevelFilter::WARN);
+        assert_eq!(dirs[0].in_span, None);
+    }
+
+    #[test]
+    fn parse_directives_empty_level() {
+        // test parse_directives with '' as log level
+        let dirs = parse_directives("crate1::mod1=wrong,crate2=");
+        assert_eq!(dirs.len(), 1, "\nparsed: {:#?}", dirs);
+        assert_eq!(dirs[0].target, Some("crate2".to_string()));
+        assert_eq!(dirs[0].level, LevelFilter::TRACE);
+        assert_eq!(dirs[0].in_span, None);
+    }
+
+    #[test]
+    fn parse_directives_global() {
+        // test parse_directives with no crate
+        let dirs = parse_directives("warn,crate2=debug");
+        assert_eq!(dirs.len(), 2, "\nparsed: {:#?}", dirs);
+        assert_eq!(dirs[0].target, None);
+        assert_eq!(dirs[0].level, LevelFilter::WARN);
+        assert_eq!(dirs[1].in_span, None);
+
+        assert_eq!(dirs[1].target, Some("crate2".to_string()));
+        assert_eq!(dirs[1].level, LevelFilter::DEBUG);
+        assert_eq!(dirs[1].in_span, None);
+    }
+
+    // helper function for tests below
+    fn test_parse_bare_level(directive_to_test: &str, level_expected: LevelFilter) {
+        let dirs = parse_directives(directive_to_test);
+        assert_eq!(
+            dirs.len(),
+            1,
+            "\ninput: \"{}\"; parsed: {:#?}",
+            directive_to_test,
+            dirs
+        );
+        assert_eq!(dirs[0].target, None);
+        assert_eq!(dirs[0].level, level_expected);
+        assert_eq!(dirs[0].in_span, None);
+    }
+
+    #[test]
+    fn parse_directives_global_bare_warn_lc() {
+        // test parse_directives with no crate, in isolation, all lowercase
+        test_parse_bare_level("warn", LevelFilter::WARN);
+    }
+
+    #[test]
+    fn parse_directives_global_bare_warn_uc() {
+        // test parse_directives with no crate, in isolation, all uppercase
+        test_parse_bare_level("WARN", LevelFilter::WARN);
+    }
+
+    #[test]
+    fn parse_directives_global_bare_warn_mixed() {
+        // test parse_directives with no crate, in isolation, mixed case
+        test_parse_bare_level("wArN", LevelFilter::WARN);
+    }
+
+    #[test]
+    fn parse_directives_valid_with_spans() {
+        let dirs = parse_directives("crate1::mod1[foo]=error,crate1::mod2[bar],crate2[baz]=debug");
+        assert_eq!(dirs.len(), 3, "\nparsed: {:#?}", dirs);
+        assert_eq!(dirs[0].target, Some("crate1::mod1".to_string()));
+        assert_eq!(dirs[0].level, LevelFilter::ERROR);
+        assert_eq!(dirs[0].in_span, Some("foo".to_string()));
+
+        assert_eq!(dirs[1].target, Some("crate1::mod2".to_string()));
+        assert_eq!(dirs[1].level, LevelFilter::TRACE);
+        assert_eq!(dirs[1].in_span, Some("bar".to_string()));
+
+        assert_eq!(dirs[2].target, Some("crate2".to_string()));
+        assert_eq!(dirs[2].level, LevelFilter::DEBUG);
+        assert_eq!(dirs[2].in_span, Some("baz".to_string()));
+    }
+
+    #[test]
+    fn parse_directives_with_dash_in_target_name() {
+        let dirs = parse_directives("target-name=info");
+        assert_eq!(dirs.len(), 1, "\nparsed: {:#?}", dirs);
+        assert_eq!(dirs[0].target, Some("target-name".to_string()));
+        assert_eq!(dirs[0].level, LevelFilter::INFO);
+        assert_eq!(dirs[0].in_span, None);
+    }
+
+    #[test]
+    fn parse_directives_with_dash_in_span_name() {
+        // Reproduces https://github.com/tokio-rs/tracing/issues/1367
+
+        let dirs = parse_directives("target[span-name]=info");
+        assert_eq!(dirs.len(), 1, "\nparsed: {:#?}", dirs);
+        assert_eq!(dirs[0].target, Some("target".to_string()));
+        assert_eq!(dirs[0].level, LevelFilter::INFO);
+        assert_eq!(dirs[0].in_span, Some("span-name".to_string()));
+    }
+
+    #[test]
+    fn parse_directives_with_special_characters_in_span_name() {
+        let span_name = "!\"#$%&'()*+-./:;<=>?@^_`|~[}";
+
+        let dirs = parse_directives(format!("target[{}]=info", span_name));
+        assert_eq!(dirs.len(), 1, "\nparsed: {:#?}", dirs);
+        assert_eq!(dirs[0].target, Some("target".to_string()));
+        assert_eq!(dirs[0].level, LevelFilter::INFO);
+        assert_eq!(dirs[0].in_span, Some(span_name.to_string()));
+    }
+
+    #[test]
+    fn parse_directives_with_invalid_span_chars() {
+        let invalid_span_name = "]{";
+
+        let dirs = parse_directives(format!("target[{}]=info", invalid_span_name));
+        assert_eq!(dirs.len(), 0, "\nparsed: {:#?}", dirs);
+    }
+}

vendor/tracing-subscriber/src/filter/env/field.rs

@@ -0,0 +1,627 @@
+use matchers::Pattern;
+use std::{
+    cmp::Ordering,
+    error::Error,
+    fmt::{self, Write},
+    str::FromStr,
+    sync::{
+        atomic::{AtomicBool, Ordering::*},
+        Arc,
+    },
+};
+
+use super::{FieldMap, LevelFilter};
+use tracing_core::field::{Field, Visit};
+
+#[derive(Clone, Debug, Eq, PartialEq)]
+pub(crate) struct Match {
+    pub(crate) name: String, // TODO: allow match patterns for names?
+    pub(crate) value: Option<ValueMatch>,
+}
+
+#[derive(Debug, Eq, PartialEq)]
+pub(crate) struct CallsiteMatch {
+    pub(crate) fields: FieldMap<ValueMatch>,
+    pub(crate) level: LevelFilter,
+}
+
+#[derive(Debug)]
+pub(crate) struct SpanMatch {
+    fields: FieldMap<(ValueMatch, AtomicBool)>,
+    level: LevelFilter,
+    has_matched: AtomicBool,
+}
+
+pub(crate) struct MatchVisitor<'a> {
+    inner: &'a SpanMatch,
+}
+
+#[derive(Debug, Clone)]
+pub(crate) enum ValueMatch {
+    /// Matches a specific `bool` value.
+    Bool(bool),
+    /// Matches a specific `f64` value.
+    F64(f64),
+    /// Matches a specific `u64` value.
+    U64(u64),
+    /// Matches a specific `i64` value.
+    I64(i64),
+    /// Matches any `NaN` `f64` value.
+    NaN,
+    /// Matches any field whose `fmt::Debug` output is equal to a fixed string.
+    Debug(MatchDebug),
+    /// Matches any field whose `fmt::Debug` output matches a regular expression
+    /// pattern.
+    Pat(Box<MatchPattern>),
+}
+
+impl Eq for ValueMatch {}
+
+impl PartialEq for ValueMatch {
+    fn eq(&self, other: &Self) -> bool {
+        use ValueMatch::*;
+        match (self, other) {
+            (Bool(a), Bool(b)) => a.eq(b),
+            (F64(a), F64(b)) => {
+                debug_assert!(!a.is_nan());
+                debug_assert!(!b.is_nan());
+
+                a.eq(b)
+            }
+            (U64(a), U64(b)) => a.eq(b),
+            (I64(a), I64(b)) => a.eq(b),
+            (NaN, NaN) => true,
+            (Pat(a), Pat(b)) => a.eq(b),
+            _ => false,
+        }
+    }
+}
+
+impl Ord for ValueMatch {
+    fn cmp(&self, other: &Self) -> Ordering {
+        use ValueMatch::*;
+        match (self, other) {
+            (Bool(this), Bool(that)) => this.cmp(that),
+            (Bool(_), _) => Ordering::Less,
+
+            (F64(this), F64(that)) => this
+                .partial_cmp(that)
+                .expect("`ValueMatch::F64` may not contain `NaN` values"),
+            (F64(_), Bool(_)) => Ordering::Greater,
+            (F64(_), _) => Ordering::Less,
+
+            (NaN, NaN) => Ordering::Equal,
+            (NaN, Bool(_)) | (NaN, F64(_)) => Ordering::Greater,
+            (NaN, _) => Ordering::Less,
+
+            (U64(this), U64(that)) => this.cmp(that),
+            (U64(_), Bool(_)) | (U64(_), F64(_)) | (U64(_), NaN) => Ordering::Greater,
+            (U64(_), _) => Ordering::Less,
+
+            (I64(this), I64(that)) => this.cmp(that),
+            (I64(_), Bool(_)) | (I64(_), F64(_)) | (I64(_), NaN) | (I64(_), U64(_)) => {
+                Ordering::Greater
+            }
+            (I64(_), _) => Ordering::Less,
+
+            (Pat(this), Pat(that)) => this.cmp(that),
+            (Pat(_), _) => Ordering::Greater,
+
+            (Debug(this), Debug(that)) => this.cmp(that),
+            (Debug(_), _) => Ordering::Greater,
+        }
+    }
+}
+
+impl PartialOrd for ValueMatch {
+    fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
+        Some(self.cmp(other))
+    }
+}
+
+/// Matches a field's `fmt::Debug` output against a regular expression pattern.
+///
+/// This is used for matching all non-literal field value filters when regular
+/// expressions are enabled.
+#[derive(Debug, Clone)]
+pub(crate) struct MatchPattern {
+    pub(crate) matcher: Pattern,
+    pattern: Arc<str>,
+}
+
+/// Matches a field's `fmt::Debug` output against a fixed string pattern.
+///
+/// This is used for matching all non-literal field value filters when regular
+/// expressions are disabled.
+#[derive(Debug, Clone)]
+pub(crate) struct MatchDebug {
+    pattern: Arc<str>,
+}
+
+/// Indicates that a field name specified in a filter directive was invalid.
+#[derive(Clone, Debug)]
+#[cfg_attr(docsrs, doc(cfg(feature = "env-filter")))]
+pub struct BadName {
+    name: String,
+}
+
+// === impl Match ===
+
+impl Match {
+    pub(crate) fn has_value(&self) -> bool {
+        self.value.is_some()
+    }
+
+    // TODO: reference count these strings?
+    pub(crate) fn name(&self) -> String {
+        self.name.clone()
+    }
+
+    pub(crate) fn parse(s: &str, regex: bool) -> Result<Self, Box<dyn Error + Send + Sync>> {
+        let mut parts = s.split('=');
+        let name = parts
+            .next()
+            .ok_or_else(|| BadName {
+                name: "".to_string(),
+            })?
+            // TODO: validate field name
+            .to_string();
+        let value = parts
+            .next()
+            .map(|part| match regex {
+                true => ValueMatch::parse_regex(part),
+                false => Ok(ValueMatch::parse_non_regex(part)),
+            })
+            .transpose()?;
+        Ok(Match { name, value })
+    }
+}
+
+impl fmt::Display for Match {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        fmt::Display::fmt(&self.name, f)?;
+        if let Some(ref value) = self.value {
+            write!(f, "={}", value)?;
+        }
+        Ok(())
+    }
+}
+
+impl Ord for Match {
+    fn cmp(&self, other: &Self) -> Ordering {
+        // Ordering for `Match` directives is based first on _whether_ a value
+        // is matched or not. This is semantically meaningful --- we would
+        // prefer to check directives that match values first as they are more
+        // specific.
+        let has_value = match (self.value.as_ref(), other.value.as_ref()) {
+            (Some(_), None) => Ordering::Greater,
+            (None, Some(_)) => Ordering::Less,
+            _ => Ordering::Equal,
+        };
+        // If both directives match a value, we fall back to the field names in
+        // length + lexicographic ordering, and if these are equal as well, we
+        // compare the match directives.
+        //
+        // This ordering is no longer semantically meaningful but is necessary
+        // so that the directives can be stored in the `BTreeMap` in a defined
+        // order.
+        has_value
+            .then_with(|| self.name.cmp(&other.name))
+            .then_with(|| self.value.cmp(&other.value))
+    }
+}
+
+impl PartialOrd for Match {
+    fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
+        Some(self.cmp(other))
+    }
+}
+
+// === impl ValueMatch ===
+
+fn value_match_f64(v: f64) -> ValueMatch {
+    if v.is_nan() {
+        ValueMatch::NaN
+    } else {
+        ValueMatch::F64(v)
+    }
+}
+
+impl ValueMatch {
+    /// Parse a `ValueMatch` that will match `fmt::Debug` fields using regular
+    /// expressions.
+    ///
+    /// This returns an error if the string didn't contain a valid `bool`,
+    /// `u64`, `i64`, or `f64` literal, and couldn't be parsed as a regular
+    /// expression.
+    #[allow(clippy::result_large_err)]
+    fn parse_regex(s: &str) -> Result<Self, matchers::BuildError> {
+        s.parse::<bool>()
+            .map(ValueMatch::Bool)
+            .or_else(|_| s.parse::<u64>().map(ValueMatch::U64))
+            .or_else(|_| s.parse::<i64>().map(ValueMatch::I64))
+            .or_else(|_| s.parse::<f64>().map(value_match_f64))
+            .or_else(|_| {
+                s.parse::<MatchPattern>()
+                    .map(|p| ValueMatch::Pat(Box::new(p)))
+            })
+    }
+
+    /// Parse a `ValueMatch` that will match `fmt::Debug` against a fixed
+    /// string.
+    ///
+    /// This does *not* return an error, because any string that isn't a valid
+    /// `bool`, `u64`, `i64`, or `f64` literal is treated as expected
+    /// `fmt::Debug` output.
+    fn parse_non_regex(s: &str) -> Self {
+        s.parse::<bool>()
+            .map(ValueMatch::Bool)
+            .or_else(|_| s.parse::<u64>().map(ValueMatch::U64))
+            .or_else(|_| s.parse::<i64>().map(ValueMatch::I64))
+            .or_else(|_| s.parse::<f64>().map(value_match_f64))
+            .unwrap_or_else(|_| ValueMatch::Debug(MatchDebug::new(s)))
+    }
+}
+
+impl fmt::Display for ValueMatch {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        match self {
+            ValueMatch::Bool(ref inner) => fmt::Display::fmt(inner, f),
+            ValueMatch::F64(ref inner) => fmt::Display::fmt(inner, f),
+            ValueMatch::NaN => fmt::Display::fmt(&f64::NAN, f),
+            ValueMatch::I64(ref inner) => fmt::Display::fmt(inner, f),
+            ValueMatch::U64(ref inner) => fmt::Display::fmt(inner, f),
+            ValueMatch::Debug(ref inner) => fmt::Display::fmt(inner, f),
+            ValueMatch::Pat(ref inner) => fmt::Display::fmt(inner, f),
+        }
+    }
+}
+
+// === impl MatchPattern ===
+
+impl FromStr for MatchPattern {
+    type Err = matchers::BuildError;
+    fn from_str(s: &str) -> Result<Self, Self::Err> {
+        let matcher = s.parse::<Pattern>()?;
+        Ok(Self {
+            matcher,
+            pattern: s.to_owned().into(),
+        })
+    }
+}
+
+impl fmt::Display for MatchPattern {
+    #[inline]
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        fmt::Display::fmt(&*self.pattern, f)
+    }
+}
+
+impl AsRef<str> for MatchPattern {
+    #[inline]
+    fn as_ref(&self) -> &str {
+        self.pattern.as_ref()
+    }
+}
+
+impl MatchPattern {
+    #[inline]
+    fn str_matches(&self, s: &impl AsRef<str>) -> bool {
+        self.matcher.matches(s)
+    }
+
+    #[inline]
+    fn debug_matches(&self, d: &impl fmt::Debug) -> bool {
+        self.matcher.debug_matches(d)
+    }
+
+    pub(super) fn into_debug_match(self) -> MatchDebug {
+        MatchDebug {
+            pattern: self.pattern,
+        }
+    }
+}
+
+impl PartialEq for MatchPattern {
+    #[inline]
+    fn eq(&self, other: &Self) -> bool {
+        self.pattern == other.pattern
+    }
+}
+
+impl Eq for MatchPattern {}
+
+impl PartialOrd for MatchPattern {
+    #[inline]
+    fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
+        Some(self.cmp(other))
+    }
+}
+
+impl Ord for MatchPattern {
+    #[inline]
+    fn cmp(&self, other: &Self) -> Ordering {
+        self.pattern.cmp(&other.pattern)
+    }
+}
+
+// === impl MatchDebug ===
+
+impl MatchDebug {
+    fn new(s: &str) -> Self {
+        Self {
+            pattern: s.to_owned().into(),
+        }
+    }
+
+    #[inline]
+    fn debug_matches(&self, d: &impl fmt::Debug) -> bool {
+        // Naively, we would probably match a value's `fmt::Debug` output by
+        // formatting it to a string, and then checking if the string is equal
+        // to the expected pattern. However, this would require allocating every
+        // time we want to match a field value against a `Debug` matcher, which
+        // can be avoided.
+        //
+        // Instead, we implement `fmt::Write` for a type that, rather than
+        // actually _writing_ the strings to something, matches them against the
+        // expected pattern, and returns an error if the pattern does not match.
+        struct Matcher<'a> {
+            pattern: &'a str,
+        }
+
+        impl fmt::Write for Matcher<'_> {
+            fn write_str(&mut self, s: &str) -> fmt::Result {
+                // If the string is longer than the remaining expected string,
+                // we know it won't match, so bail.
+                if s.len() > self.pattern.len() {
+                    return Err(fmt::Error);
+                }
+
+                // If the expected string begins with the string that was
+                // written, we are still potentially a match. Advance the
+                // position in the expected pattern to chop off the matched
+                // output, and continue.
+                if self.pattern.starts_with(s) {
+                    self.pattern = &self.pattern[s.len()..];
+                    return Ok(());
+                }
+
+                // Otherwise, the expected string doesn't include the string
+                // that was written at the current position, so the `fmt::Debug`
+                // output doesn't match! Return an error signalling that this
+                // doesn't match.
+                Err(fmt::Error)
+            }
+        }
+        let mut matcher = Matcher {
+            pattern: &self.pattern,
+        };
+
+        // Try to "write" the value's `fmt::Debug` output to a `Matcher`. This
+        // returns an error if the `fmt::Debug` implementation wrote any
+        // characters that did not match the expected pattern.
+        write!(matcher, "{:?}", d).is_ok()
+    }
+}
+
+impl fmt::Display for MatchDebug {
+    #[inline]
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        fmt::Display::fmt(&*self.pattern, f)
+    }
+}
+
+impl AsRef<str> for MatchDebug {
+    #[inline]
+    fn as_ref(&self) -> &str {
+        self.pattern.as_ref()
+    }
+}
+
+impl PartialEq for MatchDebug {
+    #[inline]
+    fn eq(&self, other: &Self) -> bool {
+        self.pattern == other.pattern
+    }
+}
+
+impl Eq for MatchDebug {}
+
+impl PartialOrd for MatchDebug {
+    #[inline]
+    fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
+        Some(self.cmp(other))
+    }
+}
+
+impl Ord for MatchDebug {
+    #[inline]
+    fn cmp(&self, other: &Self) -> Ordering {
+        self.pattern.cmp(&other.pattern)
+    }
+}
+
+// === impl BadName ===
+
+impl Error for BadName {}
+
+impl fmt::Display for BadName {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        write!(f, "invalid field name `{}`", self.name)
+    }
+}
+
+impl CallsiteMatch {
+    pub(crate) fn to_span_match(&self) -> SpanMatch {
+        let fields = self
+            .fields
+            .iter()
+            .map(|(k, v)| (k.clone(), (v.clone(), AtomicBool::new(false))))
+            .collect();
+        SpanMatch {
+            fields,
+            level: self.level,
+            has_matched: AtomicBool::new(false),
+        }
+    }
+}
+
+impl SpanMatch {
+    pub(crate) fn visitor(&self) -> MatchVisitor<'_> {
+        MatchVisitor { inner: self }
+    }
+
+    #[inline]
+    pub(crate) fn is_matched(&self) -> bool {
+        if self.has_matched.load(Acquire) {
+            return true;
+        }
+        self.is_matched_slow()
+    }
+
+    #[inline(never)]
+    fn is_matched_slow(&self) -> bool {
+        let matched = self
+            .fields
+            .values()
+            .all(|(_, matched)| matched.load(Acquire));
+        if matched {
+            self.has_matched.store(true, Release);
+        }
+        matched
+    }
+
+    #[inline]
+    pub(crate) fn filter(&self) -> Option<LevelFilter> {
+        if self.is_matched() {
+            Some(self.level)
+        } else {
+            None
+        }
+    }
+}
+
+impl Visit for MatchVisitor<'_> {
+    fn record_f64(&mut self, field: &Field, value: f64) {
+        match self.inner.fields.get(field) {
+            Some((ValueMatch::NaN, ref matched)) if value.is_nan() => {
+                matched.store(true, Release);
+            }
+            Some((ValueMatch::F64(ref e), ref matched))
+                if (value - *e).abs() < f64::EPSILON =>
+            {
+                matched.store(true, Release);
+            }
+            _ => {}
+        }
+    }
+
+    fn record_i64(&mut self, field: &Field, value: i64) {
+        use std::convert::TryInto;
+
+        match self.inner.fields.get(field) {
+            Some((ValueMatch::I64(ref e), ref matched)) if value == *e => {
+                matched.store(true, Release);
+            }
+            Some((ValueMatch::U64(ref e), ref matched)) if Ok(value) == (*e).try_into() => {
+                matched.store(true, Release);
+            }
+            _ => {}
+        }
+    }
+
+    fn record_u64(&mut self, field: &Field, value: u64) {
+        match self.inner.fields.get(field) {
+            Some((ValueMatch::U64(ref e), ref matched)) if value == *e => {
+                matched.store(true, Release);
+            }
+            _ => {}
+        }
+    }
+
+    fn record_bool(&mut self, field: &Field, value: bool) {
+        match self.inner.fields.get(field) {
+            Some((ValueMatch::Bool(ref e), ref matched)) if value == *e => {
+                matched.store(true, Release);
+            }
+            _ => {}
+        }
+    }
+
+    fn record_str(&mut self, field: &Field, value: &str) {
+        match self.inner.fields.get(field) {
+            Some((ValueMatch::Pat(ref e), ref matched)) if e.str_matches(&value) => {
+                matched.store(true, Release);
+            }
+            Some((ValueMatch::Debug(ref e), ref matched)) if e.debug_matches(&value) => {
+                matched.store(true, Release)
+            }
+            _ => {}
+        }
+    }
+
+    fn record_debug(&mut self, field: &Field, value: &dyn fmt::Debug) {
+        match self.inner.fields.get(field) {
+            Some((ValueMatch::Pat(ref e), ref matched)) if e.debug_matches(&value) => {
+                matched.store(true, Release);
+            }
+            Some((ValueMatch::Debug(ref e), ref matched)) if e.debug_matches(&value) => {
+                matched.store(true, Release)
+            }
+            _ => {}
+        }
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    #[derive(Debug)]
+    #[allow(dead_code)]
+    struct MyStruct {
+        answer: usize,
+        question: &'static str,
+    }
+
+    #[test]
+    fn debug_struct_match() {
+        let my_struct = MyStruct {
+            answer: 42,
+            question: "life, the universe, and everything",
+        };
+
+        let pattern = "MyStruct { answer: 42, question: \"life, the universe, and everything\" }";
+
+        assert_eq!(
+            format!("{:?}", my_struct),
+            pattern,
+            "`MyStruct`'s `Debug` impl doesn't output the expected string"
+        );
+
+        let matcher = MatchDebug {
+            pattern: pattern.into(),
+        };
+        assert!(matcher.debug_matches(&my_struct))
+    }
+
+    #[test]
+    fn debug_struct_not_match() {
+        let my_struct = MyStruct {
+            answer: 42,
+            question: "what shall we have for lunch?",
+        };
+
+        let pattern = "MyStruct { answer: 42, question: \"life, the universe, and everything\" }";
+
+        assert_eq!(
+            format!("{:?}", my_struct),
+            "MyStruct { answer: 42, question: \"what shall we have for lunch?\" }",
+            "`MyStruct`'s `Debug` impl doesn't output the expected string"
+        );
+
+        let matcher = MatchDebug {
+            pattern: pattern.into(),
+        };
+        assert!(!matcher.debug_matches(&my_struct))
+    }
+}

vendor/tracing-subscriber/src/filter/filter_fn.rs

@@ -0,0 +1,749 @@
+use crate::{
+    filter::LevelFilter,
+    layer::{Context, Layer},
+};
+use core::{any::type_name, fmt, marker::PhantomData};
+use tracing_core::{Interest, Metadata, Subscriber};
+
+/// A filter implemented by a closure or function pointer that
+/// determines whether a given span or event is enabled, based on its
+/// [`Metadata`].
+///
+/// This type can be used for both [per-layer filtering][plf] (using its
+/// [`Filter`] implementation) and [global filtering][global] (using its
+/// [`Layer`] implementation).
+///
+/// See the [documentation on filtering with layers][filtering] for details.
+///
+/// [`Metadata`]: tracing_core::Metadata
+/// [`Filter`]: crate::layer::Filter
+/// [`Layer`]: crate::layer::Layer
+/// [plf]: crate::layer#per-layer-filtering
+/// [global]: crate::layer#global-filtering
+/// [filtering]: crate::layer#filtering-with-layers
+#[derive(Clone)]
+pub struct FilterFn<F = fn(&Metadata<'_>) -> bool> {
+    enabled: F,
+    max_level_hint: Option<LevelFilter>,
+}
+
+/// A filter implemented by a closure or function pointer that
+/// determines whether a given span or event is enabled _dynamically_,
+/// potentially based on the current [span context].
+///
+/// This type can be used for both [per-layer filtering][plf] (using its
+/// [`Filter`] implementation) and [global filtering][global] (using its
+/// [`Layer`] implementation).
+///
+/// See the [documentation on filtering with layers][filtering] for details.
+///
+/// [span context]: crate::layer::Context
+/// [`Filter`]: crate::layer::Filter
+/// [`Layer`]: crate::layer::Layer
+/// [plf]: crate::layer#per-layer-filtering
+/// [global]: crate::layer#global-filtering
+/// [filtering]: crate::layer#filtering-with-layers
+pub struct DynFilterFn<
+    S,
+    // TODO(eliza): should these just be boxed functions?
+    F = fn(&Metadata<'_>, &Context<'_, S>) -> bool,
+    R = fn(&'static Metadata<'static>) -> Interest,
+> {
+    enabled: F,
+    register_callsite: Option<R>,
+    max_level_hint: Option<LevelFilter>,
+    _s: PhantomData<fn(S)>,
+}
+
+// === impl FilterFn ===
+
+/// Constructs a [`FilterFn`], from a function or closure that returns `true` if
+/// a span or event should be enabled, based on its [`Metadata`].
+///
+/// The returned [`FilterFn`] can be used for both [per-layer filtering][plf]
+/// (using its [`Filter`] implementation) and [global filtering][global] (using
+/// its  [`Layer`] implementation).
+///
+/// See the [documentation on filtering with layers][filtering] for details.
+///
+/// This is equivalent to calling [`FilterFn::new`].
+///
+/// [`Metadata`]: tracing_core::Metadata
+/// [`Filter`]: crate::layer::Filter
+/// [`Layer`]: crate::layer::Layer
+/// [plf]: crate::layer#per-layer-filtering
+/// [global]: crate::layer#global-filtering
+/// [filtering]: crate::layer#filtering-with-layers
+///
+/// # Examples
+///
+/// ```
+/// use tracing_subscriber::{
+///     layer::{Layer, SubscriberExt},
+///     filter,
+///     util::SubscriberInitExt,
+/// };
+///
+/// let my_filter = filter::filter_fn(|metadata| {
+///     // Only enable spans or events with the target "interesting_things"
+///     metadata.target() == "interesting_things"
+/// });
+///
+/// let my_layer = tracing_subscriber::fmt::layer();
+///
+/// tracing_subscriber::registry()
+///     .with(my_layer.with_filter(my_filter))
+///     .init();
+///
+/// // This event will not be enabled.
+/// tracing::warn!("something important but uninteresting happened!");
+///
+/// // This event will be enabled.
+/// tracing::debug!(target: "interesting_things", "an interesting minor detail...");
+/// ```
+pub fn filter_fn<F>(f: F) -> FilterFn<F>
+where
+    F: Fn(&Metadata<'_>) -> bool,
+{
+    FilterFn::new(f)
+}
+
+/// Constructs a [`DynFilterFn`] from a function or closure that returns `true`
+/// if a span or event should be enabled within a particular [span context][`Context`].
+///
+/// This is equivalent to calling [`DynFilterFn::new`].
+///
+/// Unlike [`filter_fn`], this function takes a closure or function pointer
+/// taking the [`Metadata`] for a span or event *and* the current [`Context`].
+/// This means that a [`DynFilterFn`] can choose whether to enable spans or
+/// events based on information about the _current_ span (or its parents).
+///
+/// If this is *not* necessary, use [`filter_fn`] instead.
+///
+/// The returned [`DynFilterFn`] can be used for both [per-layer filtering][plf]
+/// (using its [`Filter`] implementation) and [global filtering][global] (using
+/// its  [`Layer`] implementation).
+///
+/// See the [documentation on filtering with layers][filtering] for details.
+///
+/// # Examples
+///
+/// ```
+/// use tracing_subscriber::{
+///     layer::{Layer, SubscriberExt},
+///     filter,
+///     util::SubscriberInitExt,
+/// };
+///
+/// // Only enable spans or events within a span named "interesting_span".
+/// let my_filter = filter::dynamic_filter_fn(|metadata, cx| {
+///     // If this *is* "interesting_span", make sure to enable it.
+///     if metadata.is_span() && metadata.name() == "interesting_span" {
+///         return true;
+///     }
+///
+///     // Otherwise, are we in an interesting span?
+///     if let Some(current_span) = cx.lookup_current() {
+///         return current_span.name() == "interesting_span";
+///     }
+///
+///     false
+/// });
+///
+/// let my_layer = tracing_subscriber::fmt::layer();
+///
+/// tracing_subscriber::registry()
+///     .with(my_layer.with_filter(my_filter))
+///     .init();
+///
+/// // This event will not be enabled.
+/// tracing::info!("something happened");
+///
+/// tracing::info_span!("interesting_span").in_scope(|| {
+///     // This event will be enabled.
+///     tracing::debug!("something else happened");
+/// });
+/// ```
+///
+/// [`Filter`]: crate::layer::Filter
+/// [`Layer`]: crate::layer::Layer
+/// [plf]: crate::layer#per-layer-filtering
+/// [global]: crate::layer#global-filtering
+/// [filtering]: crate::layer#filtering-with-layers
+/// [`Context`]: crate::layer::Context
+/// [`Metadata`]: tracing_core::Metadata
+pub fn dynamic_filter_fn<S, F>(f: F) -> DynFilterFn<S, F>
+where
+    F: Fn(&Metadata<'_>, &Context<'_, S>) -> bool,
+{
+    DynFilterFn::new(f)
+}
+
+impl<F> FilterFn<F>
+where
+    F: Fn(&Metadata<'_>) -> bool,
+{
+    /// Constructs a [`FilterFn`] from a function or closure that returns `true`
+    /// if a span or event should be enabled, based on its [`Metadata`].
+    ///
+    /// If determining whether a span or event should be enabled also requires
+    /// information about the current span context, use [`DynFilterFn`] instead.
+    ///
+    /// See the [documentation on per-layer filtering][plf] for details on using
+    /// [`Filter`]s.
+    ///
+    /// [`Filter`]: crate::layer::Filter
+    /// [plf]: crate::layer#per-layer-filtering
+    /// [`Metadata`]: tracing_core::Metadata
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// use tracing_subscriber::{
+    ///     layer::{Layer, SubscriberExt},
+    ///     filter::FilterFn,
+    ///     util::SubscriberInitExt,
+    /// };
+    ///
+    /// let my_filter = FilterFn::new(|metadata| {
+    ///     // Only enable spans or events with the target "interesting_things"
+    ///     metadata.target() == "interesting_things"
+    /// });
+    ///
+    /// let my_layer = tracing_subscriber::fmt::layer();
+    ///
+    /// tracing_subscriber::registry()
+    ///     .with(my_layer.with_filter(my_filter))
+    ///     .init();
+    ///
+    /// // This event will not be enabled.
+    /// tracing::warn!("something important but uninteresting happened!");
+    ///
+    /// // This event will be enabled.
+    /// tracing::debug!(target: "interesting_things", "an interesting minor detail...");
+    /// ```
+    pub fn new(enabled: F) -> Self {
+        Self {
+            enabled,
+            max_level_hint: None,
+        }
+    }
+
+    /// Sets the highest verbosity [`Level`] the filter function will enable.
+    ///
+    /// The value passed to this method will be returned by this `FilterFn`'s
+    /// [`Filter::max_level_hint`] method.
+    ///
+    /// If the provided function will not enable all levels, it is recommended
+    /// to call this method to configure it with the most verbose level it will
+    /// enable.
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// use tracing_subscriber::{
+    ///     layer::{Layer, SubscriberExt},
+    ///     filter::{filter_fn, LevelFilter},
+    ///     util::SubscriberInitExt,
+    /// };
+    /// use tracing_core::Level;
+    ///
+    /// let my_filter = filter_fn(|metadata| {
+    ///     // Only enable spans or events with targets starting with `my_crate`
+    ///     // and levels at or below `INFO`.
+    ///     metadata.level() <= &Level::INFO && metadata.target().starts_with("my_crate")
+    /// })
+    ///     // Since the filter closure will only enable the `INFO` level and
+    ///     // below, set the max level hint
+    ///     .with_max_level_hint(LevelFilter::INFO);
+    ///
+    /// let my_layer = tracing_subscriber::fmt::layer();
+    ///
+    /// tracing_subscriber::registry()
+    ///     .with(my_layer.with_filter(my_filter))
+    ///     .init();
+    /// ```
+    ///
+    /// [`Level`]: tracing_core::Level
+    /// [`Filter::max_level_hint`]: crate::layer::Filter::max_level_hint
+    pub fn with_max_level_hint(self, max_level_hint: impl Into<LevelFilter>) -> Self {
+        Self {
+            max_level_hint: Some(max_level_hint.into()),
+            ..self
+        }
+    }
+
+    #[inline]
+    pub(in crate::filter) fn is_enabled(&self, metadata: &Metadata<'_>) -> bool {
+        let enabled = (self.enabled)(metadata);
+        debug_assert!(
+            !enabled || self.is_below_max_level(metadata),
+            "FilterFn<{}> claimed it would only enable {:?} and below, \
+            but it enabled metadata with the {:?} level\nmetadata={:#?}",
+            type_name::<F>(),
+            self.max_level_hint.unwrap(),
+            metadata.level(),
+            metadata,
+        );
+
+        enabled
+    }
+
+    #[inline]
+    pub(in crate::filter) fn is_callsite_enabled(
+        &self,
+        metadata: &'static Metadata<'static>,
+    ) -> Interest {
+        // Because `self.enabled` takes a `Metadata` only (and no `Context`
+        // parameter), we can reasonably assume its results are cacheable, and
+        // just return `Interest::always`/`Interest::never`.
+        if (self.enabled)(metadata) {
+            debug_assert!(
+                self.is_below_max_level(metadata),
+                "FilterFn<{}> claimed it was only interested in {:?} and below, \
+                but it enabled metadata with the {:?} level\nmetadata={:#?}",
+                type_name::<F>(),
+                self.max_level_hint.unwrap(),
+                metadata.level(),
+                metadata,
+            );
+            return Interest::always();
+        }
+
+        Interest::never()
+    }
+
+    fn is_below_max_level(&self, metadata: &Metadata<'_>) -> bool {
+        self.max_level_hint
+            .as_ref()
+            .map(|hint| metadata.level() <= hint)
+            .unwrap_or(true)
+    }
+}
+
+impl<S, F> Layer<S> for FilterFn<F>
+where
+    F: Fn(&Metadata<'_>) -> bool + 'static,
+    S: Subscriber,
+{
+    fn enabled(&self, metadata: &Metadata<'_>, _: Context<'_, S>) -> bool {
+        self.is_enabled(metadata)
+    }
+
+    fn register_callsite(&self, metadata: &'static Metadata<'static>) -> Interest {
+        self.is_callsite_enabled(metadata)
+    }
+
+    fn max_level_hint(&self) -> Option<LevelFilter> {
+        self.max_level_hint
+    }
+}
+
+impl<F> From<F> for FilterFn<F>
+where
+    F: Fn(&Metadata<'_>) -> bool,
+{
+    fn from(enabled: F) -> Self {
+        Self::new(enabled)
+    }
+}
+
+impl<F> fmt::Debug for FilterFn<F> {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        f.debug_struct("FilterFn")
+            .field("enabled", &format_args!("{}", type_name::<F>()))
+            .field("max_level_hint", &self.max_level_hint)
+            .finish()
+    }
+}
+
+// === impl DynFilterFn ==
+
+impl<S, F> DynFilterFn<S, F>
+where
+    F: Fn(&Metadata<'_>, &Context<'_, S>) -> bool,
+{
+    /// Constructs a [`Filter`] from a function or closure that returns `true`
+    /// if a span or event should be enabled in the current [span
+    /// context][`Context`].
+    ///
+    /// Unlike [`FilterFn`], a `DynFilterFn` is constructed from a closure or
+    /// function pointer that takes both the [`Metadata`] for a span or event
+    /// *and* the current [`Context`]. This means that a [`DynFilterFn`] can
+    /// choose whether to enable spans or events based on information about the
+    /// _current_ span (or its parents).
+    ///
+    /// If this is *not* necessary, use [`FilterFn`] instead.
+    ///
+    /// See the [documentation on per-layer filtering][plf] for details on using
+    /// [`Filter`]s.
+    ///
+    /// [`Filter`]: crate::layer::Filter
+    /// [plf]: crate::layer#per-layer-filtering
+    /// [`Context`]: crate::layer::Context
+    /// [`Metadata`]: tracing_core::Metadata
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// use tracing_subscriber::{
+    ///     layer::{Layer, SubscriberExt},
+    ///     filter::DynFilterFn,
+    ///     util::SubscriberInitExt,
+    /// };
+    ///
+    /// // Only enable spans or events within a span named "interesting_span".
+    /// let my_filter = DynFilterFn::new(|metadata, cx| {
+    ///     // If this *is* "interesting_span", make sure to enable it.
+    ///     if metadata.is_span() && metadata.name() == "interesting_span" {
+    ///         return true;
+    ///     }
+    ///
+    ///     // Otherwise, are we in an interesting span?
+    ///     if let Some(current_span) = cx.lookup_current() {
+    ///         return current_span.name() == "interesting_span";
+    ///     }
+    ///
+    ///     false
+    /// });
+    ///
+    /// let my_layer = tracing_subscriber::fmt::layer();
+    ///
+    /// tracing_subscriber::registry()
+    ///     .with(my_layer.with_filter(my_filter))
+    ///     .init();
+    ///
+    /// // This event will not be enabled.
+    /// tracing::info!("something happened");
+    ///
+    /// tracing::info_span!("interesting_span").in_scope(|| {
+    ///     // This event will be enabled.
+    ///     tracing::debug!("something else happened");
+    /// });
+    /// ```
+    pub fn new(enabled: F) -> Self {
+        Self {
+            enabled,
+            register_callsite: None,
+            max_level_hint: None,
+            _s: PhantomData,
+        }
+    }
+}
+
+impl<S, F, R> DynFilterFn<S, F, R>
+where
+    F: Fn(&Metadata<'_>, &Context<'_, S>) -> bool,
+{
+    /// Sets the highest verbosity [`Level`] the filter function will enable.
+    ///
+    /// The value passed to this method will be returned by this `DynFilterFn`'s
+    /// [`Filter::max_level_hint`] method.
+    ///
+    /// If the provided function will not enable all levels, it is recommended
+    /// to call this method to configure it with the most verbose level it will
+    /// enable.
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// use tracing_subscriber::{
+    ///     layer::{Layer, SubscriberExt},
+    ///     filter::{DynFilterFn, LevelFilter},
+    ///     util::SubscriberInitExt,
+    /// };
+    /// use tracing_core::Level;
+    ///
+    /// // Only enable spans or events with levels at or below `INFO`, if
+    /// // we are inside a span called "interesting_span".
+    /// let my_filter = DynFilterFn::new(|metadata, cx| {
+    ///     // If the level is greater than INFO, disable it.
+    ///     if metadata.level() > &Level::INFO {
+    ///         return false;
+    ///     }
+    ///
+    ///     // If any span in the current scope is named "interesting_span",
+    ///     // enable this span or event.
+    ///     for span in cx.lookup_current().iter().flat_map(|span| span.scope()) {
+    ///         if span.name() == "interesting_span" {
+    ///             return true;
+    ///          }
+    ///     }
+    ///
+    ///     // Otherwise, disable it.
+    ///     false
+    /// })
+    ///     // Since the filter closure will only enable the `INFO` level and
+    ///     // below, set the max level hint
+    ///     .with_max_level_hint(LevelFilter::INFO);
+    ///
+    /// let my_layer = tracing_subscriber::fmt::layer();
+    ///
+    /// tracing_subscriber::registry()
+    ///     .with(my_layer.with_filter(my_filter))
+    ///     .init();
+    /// ```
+    ///
+    /// [`Level`]: tracing_core::Level
+    /// [`Filter::max_level_hint`]: crate::layer::Filter::max_level_hint
+    pub fn with_max_level_hint(self, max_level_hint: impl Into<LevelFilter>) -> Self {
+        Self {
+            max_level_hint: Some(max_level_hint.into()),
+            ..self
+        }
+    }
+
+    /// Adds a function for filtering callsites to this filter.
+    ///
+    /// When this filter's [`Filter::callsite_enabled`][cse] method is called,
+    /// the provided function will be used rather than the default.
+    ///
+    /// By default, `DynFilterFn` assumes that, because the filter _may_ depend
+    /// dynamically on the current [span context], its result should never be
+    /// cached. However, some filtering strategies may require dynamic information
+    /// from the current span context in *some* cases, but are able to make
+    /// static filtering decisions from [`Metadata`] alone in others.
+    ///
+    /// For example, consider the filter given in the example for
+    /// [`DynFilterFn::new`]. That filter enables all spans named
+    /// "interesting_span", and any events and spans that occur inside of an
+    /// interesting span. Since the span's name is part of its static
+    /// [`Metadata`], the "interesting_span" can be enabled in
+    /// [`callsite_enabled`][cse]:
+    ///
+    /// ```
+    /// use tracing_subscriber::{
+    ///     layer::{Layer, SubscriberExt},
+    ///     filter::DynFilterFn,
+    ///     util::SubscriberInitExt,
+    /// };
+    /// use tracing_core::subscriber::Interest;
+    ///
+    /// // Only enable spans or events within a span named "interesting_span".
+    /// let my_filter = DynFilterFn::new(|metadata, cx| {
+    ///     // If this *is* "interesting_span", make sure to enable it.
+    ///     if metadata.is_span() && metadata.name() == "interesting_span" {
+    ///         return true;
+    ///     }
+    ///
+    ///     // Otherwise, are we in an interesting span?
+    ///     if let Some(current_span) = cx.lookup_current() {
+    ///         return current_span.name() == "interesting_span";
+    ///     }
+    ///
+    ///     false
+    /// }).with_callsite_filter(|metadata| {
+    ///     // If this is an "interesting_span", we know we will always
+    ///     // enable it.
+    ///     if metadata.is_span() && metadata.name() == "interesting_span" {
+    ///         return Interest::always();
+    ///     }
+    ///
+    ///     // Otherwise, it depends on whether or not we're in an interesting
+    ///     // span. You'll have to ask us again for each span/event!
+    ///     Interest::sometimes()
+    /// });
+    ///
+    /// let my_layer = tracing_subscriber::fmt::layer();
+    ///
+    /// tracing_subscriber::registry()
+    ///     .with(my_layer.with_filter(my_filter))
+    ///     .init();
+    /// ```
+    ///
+    /// [cse]: crate::layer::Filter::callsite_enabled
+    /// [`enabled`]: crate::layer::Filter::enabled
+    /// [`Metadata`]: tracing_core::Metadata
+    /// [span context]: crate::layer::Context
+    pub fn with_callsite_filter<R2>(self, callsite_enabled: R2) -> DynFilterFn<S, F, R2>
+    where
+        R2: Fn(&'static Metadata<'static>) -> Interest,
+    {
+        let register_callsite = Some(callsite_enabled);
+        let DynFilterFn {
+            enabled,
+            max_level_hint,
+            _s,
+            ..
+        } = self;
+        DynFilterFn {
+            enabled,
+            register_callsite,
+            max_level_hint,
+            _s,
+        }
+    }
+
+    fn default_callsite_enabled(&self, metadata: &Metadata<'_>) -> Interest {
+        // If it's below the configured max level, assume that `enabled` will
+        // never enable it...
+        if !is_below_max_level(&self.max_level_hint, metadata) {
+            debug_assert!(
+                !(self.enabled)(metadata, &Context::none()),
+                "DynFilterFn<{}> claimed it would only enable {:?} and below, \
+                but it enabled metadata with the {:?} level\nmetadata={:#?}",
+                type_name::<F>(),
+                self.max_level_hint.unwrap(),
+                metadata.level(),
+                metadata,
+            );
+            return Interest::never();
+        }
+
+        // Otherwise, since this `enabled` function is dynamic and depends on
+        // the current context, we don't know whether this span or event will be
+        // enabled or not. Ask again every time it's recorded!
+        Interest::sometimes()
+    }
+}
+
+impl<S, F, R> DynFilterFn<S, F, R>
+where
+    F: Fn(&Metadata<'_>, &Context<'_, S>) -> bool,
+    R: Fn(&'static Metadata<'static>) -> Interest,
+{
+    #[inline]
+    fn is_enabled(&self, metadata: &Metadata<'_>, cx: &Context<'_, S>) -> bool {
+        let enabled = (self.enabled)(metadata, cx);
+        debug_assert!(
+            !enabled || is_below_max_level(&self.max_level_hint, metadata),
+            "DynFilterFn<{}> claimed it would only enable {:?} and below, \
+            but it enabled metadata with the {:?} level\nmetadata={:#?}",
+            type_name::<F>(),
+            self.max_level_hint.unwrap(),
+            metadata.level(),
+            metadata,
+        );
+
+        enabled
+    }
+
+    #[inline]
+    fn is_callsite_enabled(&self, metadata: &'static Metadata<'static>) -> Interest {
+        let interest = self
+            .register_callsite
+            .as_ref()
+            .map(|callsite_enabled| callsite_enabled(metadata))
+            .unwrap_or_else(|| self.default_callsite_enabled(metadata));
+        debug_assert!(
+            interest.is_never() || is_below_max_level(&self.max_level_hint, metadata),
+            "DynFilterFn<{}, {}> claimed it was only interested in {:?} and below, \
+            but it enabled metadata with the {:?} level\nmetadata={:#?}",
+            type_name::<F>(),
+            type_name::<R>(),
+            self.max_level_hint.unwrap(),
+            metadata.level(),
+            metadata,
+        );
+
+        interest
+    }
+}
+
+impl<S, F, R> Layer<S> for DynFilterFn<S, F, R>
+where
+    F: Fn(&Metadata<'_>, &Context<'_, S>) -> bool + 'static,
+    R: Fn(&'static Metadata<'static>) -> Interest + 'static,
+    S: Subscriber,
+{
+    fn enabled(&self, metadata: &Metadata<'_>, cx: Context<'_, S>) -> bool {
+        self.is_enabled(metadata, &cx)
+    }
+
+    fn register_callsite(&self, metadata: &'static Metadata<'static>) -> Interest {
+        self.is_callsite_enabled(metadata)
+    }
+
+    fn max_level_hint(&self) -> Option<LevelFilter> {
+        self.max_level_hint
+    }
+}
+
+impl<S, F, R> fmt::Debug for DynFilterFn<S, F, R> {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        let mut s = f.debug_struct("DynFilterFn");
+        s.field("enabled", &format_args!("{}", type_name::<F>()));
+        if self.register_callsite.is_some() {
+            s.field(
+                "register_callsite",
+                &format_args!("Some({})", type_name::<R>()),
+            );
+        } else {
+            s.field("register_callsite", &format_args!("None"));
+        }
+
+        s.field("max_level_hint", &self.max_level_hint).finish()
+    }
+}
+
+impl<S, F, R> Clone for DynFilterFn<S, F, R>
+where
+    F: Clone,
+    R: Clone,
+{
+    fn clone(&self) -> Self {
+        Self {
+            enabled: self.enabled.clone(),
+            register_callsite: self.register_callsite.clone(),
+            max_level_hint: self.max_level_hint,
+            _s: PhantomData,
+        }
+    }
+}
+
+impl<F, S> From<F> for DynFilterFn<S, F>
+where
+    F: Fn(&Metadata<'_>, &Context<'_, S>) -> bool,
+{
+    fn from(f: F) -> Self {
+        Self::new(f)
+    }
+}
+
+// === PLF impls ===
+
+feature! {
+    #![all(feature = "registry", feature = "std")]
+    use crate::layer::Filter;
+
+    impl<S, F> Filter<S> for FilterFn<F>
+    where
+        F: Fn(&Metadata<'_>) -> bool,
+    {
+        fn enabled(&self, metadata: &Metadata<'_>, _: &Context<'_, S>) -> bool {
+            self.is_enabled(metadata)
+        }
+
+        fn callsite_enabled(&self, metadata: &'static Metadata<'static>) -> Interest {
+            self.is_callsite_enabled(metadata)
+        }
+
+        fn max_level_hint(&self) -> Option<LevelFilter> {
+            self.max_level_hint
+        }
+    }
+
+    impl<S, F, R> Filter<S> for DynFilterFn<S, F, R>
+    where
+        F: Fn(&Metadata<'_>, &Context<'_, S>) -> bool,
+        R: Fn(&'static Metadata<'static>) -> Interest,
+    {
+        fn enabled(&self, metadata: &Metadata<'_>, cx: &Context<'_, S>) -> bool {
+            self.is_enabled(metadata, cx)
+        }
+
+        fn callsite_enabled(&self, metadata: &'static Metadata<'static>) -> Interest {
+            self.is_callsite_enabled(metadata)
+        }
+
+        fn max_level_hint(&self) -> Option<LevelFilter> {
+            self.max_level_hint
+        }
+    }
+}
+
+fn is_below_max_level(hint: &Option<LevelFilter>, metadata: &Metadata<'_>) -> bool {
+    hint.as_ref()
+        .map(|hint| metadata.level() <= hint)
+        .unwrap_or(true)
+}

vendor/tracing-subscriber/src/filter/layer_filters/combinator.rs

@@ -0,0 +1,542 @@
+//! Filter combinators
+use crate::layer::{Context, Filter};
+use std::{cmp, fmt, marker::PhantomData};
+use tracing_core::{
+    span::{Attributes, Id, Record},
+    subscriber::Interest,
+    LevelFilter, Metadata,
+};
+
+/// Combines two [`Filter`]s so that spans and events are enabled if and only if
+/// *both* filters return `true`.
+///
+/// This type is typically returned by the [`FilterExt::and`] method. See that
+/// method's documentation for details.
+///
+/// [`Filter`]: crate::layer::Filter
+/// [`FilterExt::and`]: crate::filter::FilterExt::and
+pub struct And<A, B, S> {
+    a: A,
+    b: B,
+    _s: PhantomData<fn(S)>,
+}
+
+/// Combines two [`Filter`]s so that spans and events are enabled if *either* filter
+/// returns `true`.
+///
+/// This type is typically returned by the [`FilterExt::or`] method. See that
+/// method's documentation for details.
+///
+/// [`Filter`]: crate::layer::Filter
+/// [`FilterExt::or`]: crate::filter::FilterExt::or
+pub struct Or<A, B, S> {
+    a: A,
+    b: B,
+    _s: PhantomData<fn(S)>,
+}
+
+/// Inverts the result of a [`Filter`].
+///
+/// If the wrapped filter would enable a span or event, it will be disabled. If
+/// it would disable a span or event, that span or event will be enabled.
+///
+/// This type is typically returned by the [`FilterExt::not`] method. See that
+/// method's documentation for details.
+///
+/// [`Filter`]: crate::layer::Filter
+/// [`FilterExt::not`]: crate::filter::FilterExt::not
+pub struct Not<A, S> {
+    a: A,
+    _s: PhantomData<fn(S)>,
+}
+
+// === impl And ===
+
+impl<A, B, S> And<A, B, S>
+where
+    A: Filter<S>,
+    B: Filter<S>,
+{
+    /// Combines two [`Filter`]s so that spans and events are enabled if and only if
+    /// *both* filters return `true`.
+    ///
+    /// # Examples
+    ///
+    /// Enabling spans or events if they have both a particular target *and* are
+    /// above a certain level:
+    ///
+    /// ```ignore
+    /// use tracing_subscriber::{
+    ///     filter::{filter_fn, LevelFilter, combinator::And},
+    ///     prelude::*,
+    /// };
+    ///
+    /// // Enables spans and events with targets starting with `interesting_target`:
+    /// let target_filter = filter_fn(|meta| {
+    ///     meta.target().starts_with("interesting_target")
+    /// });
+    ///
+    /// // Enables spans and events with levels `INFO` and below:
+    /// let level_filter = LevelFilter::INFO;
+    ///
+    /// // Combine the two filters together so that a span or event is only enabled
+    /// // if *both* filters would enable it:
+    /// let filter = And::new(level_filter, target_filter);
+    ///
+    /// tracing_subscriber::registry()
+    ///     .with(tracing_subscriber::fmt::layer().with_filter(filter))
+    ///     .init();
+    ///
+    /// // This event will *not* be enabled:
+    /// tracing::info!("an event with an uninteresting target");
+    ///
+    /// // This event *will* be enabled:
+    /// tracing::info!(target: "interesting_target", "a very interesting event");
+    ///
+    /// // This event will *not* be enabled:
+    /// tracing::debug!(target: "interesting_target", "interesting debug event...");
+    /// ```
+    ///
+    /// [`Filter`]: crate::layer::Filter
+    pub(crate) fn new(a: A, b: B) -> Self {
+        Self {
+            a,
+            b,
+            _s: PhantomData,
+        }
+    }
+}
+
+impl<A, B, S> Filter<S> for And<A, B, S>
+where
+    A: Filter<S>,
+    B: Filter<S>,
+{
+    #[inline]
+    fn enabled(&self, meta: &Metadata<'_>, cx: &Context<'_, S>) -> bool {
+        self.a.enabled(meta, cx) && self.b.enabled(meta, cx)
+    }
+
+    fn callsite_enabled(&self, meta: &'static Metadata<'static>) -> Interest {
+        let a = self.a.callsite_enabled(meta);
+        if a.is_never() {
+            return a;
+        }
+
+        let b = self.b.callsite_enabled(meta);
+
+        if !b.is_always() {
+            return b;
+        }
+
+        a
+    }
+
+    fn max_level_hint(&self) -> Option<LevelFilter> {
+        // If either hint is `None`, return `None`. Otherwise, return the most restrictive.
+        cmp::min(self.a.max_level_hint(), self.b.max_level_hint())
+    }
+
+    #[inline]
+    fn event_enabled(&self, event: &tracing_core::Event<'_>, cx: &Context<'_, S>) -> bool {
+        self.a.event_enabled(event, cx) && self.b.event_enabled(event, cx)
+    }
+
+    #[inline]
+    fn on_new_span(&self, attrs: &Attributes<'_>, id: &Id, ctx: Context<'_, S>) {
+        self.a.on_new_span(attrs, id, ctx.clone());
+        self.b.on_new_span(attrs, id, ctx)
+    }
+
+    #[inline]
+    fn on_record(&self, id: &Id, values: &Record<'_>, ctx: Context<'_, S>) {
+        self.a.on_record(id, values, ctx.clone());
+        self.b.on_record(id, values, ctx);
+    }
+
+    #[inline]
+    fn on_enter(&self, id: &Id, ctx: Context<'_, S>) {
+        self.a.on_enter(id, ctx.clone());
+        self.b.on_enter(id, ctx);
+    }
+
+    #[inline]
+    fn on_exit(&self, id: &Id, ctx: Context<'_, S>) {
+        self.a.on_exit(id, ctx.clone());
+        self.b.on_exit(id, ctx);
+    }
+
+    #[inline]
+    fn on_close(&self, id: Id, ctx: Context<'_, S>) {
+        self.a.on_close(id.clone(), ctx.clone());
+        self.b.on_close(id, ctx);
+    }
+}
+
+impl<A, B, S> Clone for And<A, B, S>
+where
+    A: Clone,
+    B: Clone,
+{
+    fn clone(&self) -> Self {
+        Self {
+            a: self.a.clone(),
+            b: self.b.clone(),
+            _s: PhantomData,
+        }
+    }
+}
+
+impl<A, B, S> fmt::Debug for And<A, B, S>
+where
+    A: fmt::Debug,
+    B: fmt::Debug,
+{
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        f.debug_struct("And")
+            .field("a", &self.a)
+            .field("b", &self.b)
+            .finish()
+    }
+}
+
+// === impl Or ===
+
+impl<A, B, S> Or<A, B, S>
+where
+    A: Filter<S>,
+    B: Filter<S>,
+{
+    /// Combines two [`Filter`]s so that spans and events are enabled if *either* filter
+    /// returns `true`.
+    ///
+    /// # Examples
+    ///
+    /// Enabling spans and events at the `INFO` level and above, and all spans
+    /// and events with a particular target:
+    ///
+    /// ```ignore
+    /// use tracing_subscriber::{
+    ///     filter::{filter_fn, LevelFilter, combinator::Or},
+    ///     prelude::*,
+    /// };
+    ///
+    /// // Enables spans and events with targets starting with `interesting_target`:
+    /// let target_filter = filter_fn(|meta| {
+    ///     meta.target().starts_with("interesting_target")
+    /// });
+    ///
+    /// // Enables spans and events with levels `INFO` and below:
+    /// let level_filter = LevelFilter::INFO;
+    ///
+    /// // Combine the two filters together so that a span or event is enabled
+    /// // if it is at INFO or lower, or if it has a target starting with
+    /// // `interesting_target`.
+    /// let filter = Or::new(level_filter, target_filter);
+    ///
+    /// tracing_subscriber::registry()
+    ///     .with(tracing_subscriber::fmt::layer().with_filter(filter))
+    ///     .init();
+    ///
+    /// // This event will *not* be enabled:
+    /// tracing::debug!("an uninteresting event");
+    ///
+    /// // This event *will* be enabled:
+    /// tracing::info!("an uninteresting INFO event");
+    ///
+    /// // This event *will* be enabled:
+    /// tracing::info!(target: "interesting_target", "a very interesting event");
+    ///
+    /// // This event *will* be enabled:
+    /// tracing::debug!(target: "interesting_target", "interesting debug event...");
+    /// ```
+    ///
+    /// Enabling a higher level for a particular target by using `Or` in
+    /// conjunction with the [`And`] combinator:
+    ///
+    /// ```ignore
+    /// use tracing_subscriber::{
+    ///     filter::{filter_fn, LevelFilter, combinator},
+    ///     prelude::*,
+    /// };
+    ///
+    /// // This filter will enable spans and events with targets beginning with
+    /// // `my_crate`:
+    /// let my_crate = filter_fn(|meta| {
+    ///     meta.target().starts_with("my_crate")
+    /// });
+    ///
+    /// // Combine the `my_crate` filter with a `LevelFilter` to produce a filter
+    /// // that will enable the `INFO` level and lower for spans and events with
+    /// // `my_crate` targets:
+    /// let filter = combinator::And::new(my_crate, LevelFilter::INFO);
+    ///
+    /// // If a span or event *doesn't* have a target beginning with
+    /// // `my_crate`, enable it if it has the `WARN` level or lower:
+    /// // let filter = combinator::Or::new(filter, LevelFilter::WARN);
+    ///
+    /// tracing_subscriber::registry()
+    ///     .with(tracing_subscriber::fmt::layer().with_filter(filter))
+    ///     .init();
+    /// ```
+    ///
+    /// [`Filter`]: crate::layer::Filter
+    pub(crate) fn new(a: A, b: B) -> Self {
+        Self {
+            a,
+            b,
+            _s: PhantomData,
+        }
+    }
+}
+
+impl<A, B, S> Filter<S> for Or<A, B, S>
+where
+    A: Filter<S>,
+    B: Filter<S>,
+{
+    #[inline]
+    fn enabled(&self, meta: &Metadata<'_>, cx: &Context<'_, S>) -> bool {
+        self.a.enabled(meta, cx) || self.b.enabled(meta, cx)
+    }
+
+    fn callsite_enabled(&self, meta: &'static Metadata<'static>) -> Interest {
+        let a = self.a.callsite_enabled(meta);
+        let b = self.b.callsite_enabled(meta);
+
+        // If either filter will always enable the span or event, return `always`.
+        if a.is_always() || b.is_always() {
+            return Interest::always();
+        }
+
+        // Okay, if either filter will sometimes enable the span or event,
+        // return `sometimes`.
+        if a.is_sometimes() || b.is_sometimes() {
+            return Interest::sometimes();
+        }
+
+        debug_assert!(
+            a.is_never() && b.is_never(),
+            "if neither filter was `always` or `sometimes`, both must be `never` (a={:?}; b={:?})",
+            a,
+            b,
+        );
+        Interest::never()
+    }
+
+    fn max_level_hint(&self) -> Option<LevelFilter> {
+        // If either hint is `None`, return `None`. Otherwise, return the less restrictive.
+        Some(cmp::max(self.a.max_level_hint()?, self.b.max_level_hint()?))
+    }
+
+    #[inline]
+    fn event_enabled(&self, event: &tracing_core::Event<'_>, cx: &Context<'_, S>) -> bool {
+        self.a.event_enabled(event, cx) || self.b.event_enabled(event, cx)
+    }
+
+    #[inline]
+    fn on_new_span(&self, attrs: &Attributes<'_>, id: &Id, ctx: Context<'_, S>) {
+        self.a.on_new_span(attrs, id, ctx.clone());
+        self.b.on_new_span(attrs, id, ctx)
+    }
+
+    #[inline]
+    fn on_record(&self, id: &Id, values: &Record<'_>, ctx: Context<'_, S>) {
+        self.a.on_record(id, values, ctx.clone());
+        self.b.on_record(id, values, ctx);
+    }
+
+    #[inline]
+    fn on_enter(&self, id: &Id, ctx: Context<'_, S>) {
+        self.a.on_enter(id, ctx.clone());
+        self.b.on_enter(id, ctx);
+    }
+
+    #[inline]
+    fn on_exit(&self, id: &Id, ctx: Context<'_, S>) {
+        self.a.on_exit(id, ctx.clone());
+        self.b.on_exit(id, ctx);
+    }
+
+    #[inline]
+    fn on_close(&self, id: Id, ctx: Context<'_, S>) {
+        self.a.on_close(id.clone(), ctx.clone());
+        self.b.on_close(id, ctx);
+    }
+}
+
+impl<A, B, S> Clone for Or<A, B, S>
+where
+    A: Clone,
+    B: Clone,
+{
+    fn clone(&self) -> Self {
+        Self {
+            a: self.a.clone(),
+            b: self.b.clone(),
+            _s: PhantomData,
+        }
+    }
+}
+
+impl<A, B, S> fmt::Debug for Or<A, B, S>
+where
+    A: fmt::Debug,
+    B: fmt::Debug,
+{
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        f.debug_struct("Or")
+            .field("a", &self.a)
+            .field("b", &self.b)
+            .finish()
+    }
+}
+
+// === impl Not ===
+
+impl<A, S> Not<A, S>
+where
+    A: Filter<S>,
+{
+    /// Inverts the result of a [`Filter`].
+    ///
+    /// If the wrapped filter would enable a span or event, it will be disabled. If
+    /// it would disable a span or event, that span or event will be enabled.
+    ///
+    /// This inverts the values returned by the [`enabled`] and [`callsite_enabled`]
+    /// methods on the wrapped filter; it does *not* invert [`event_enabled`], as
+    /// filters which do not implement filtering on event field values will return
+    /// the default `true` even for events that their [`enabled`] method disables.
+    ///
+    /// Consider a normal filter defined as:
+    ///
+    /// ```ignore (pseudo-code)
+    /// // for spans
+    /// match callsite_enabled() {
+    ///     ALWAYS => on_span(),
+    ///     SOMETIMES => if enabled() { on_span() },
+    ///     NEVER => (),
+    /// }
+    /// // for events
+    /// match callsite_enabled() {
+    ///    ALWAYS => on_event(),
+    ///    SOMETIMES => if enabled() && event_enabled() { on_event() },
+    ///    NEVER => (),
+    /// }
+    /// ```
+    ///
+    /// and an inverted filter defined as:
+    ///
+    /// ```ignore (pseudo-code)
+    /// // for spans
+    /// match callsite_enabled() {
+    ///     ALWAYS => (),
+    ///     SOMETIMES => if !enabled() { on_span() },
+    ///     NEVER => on_span(),
+    /// }
+    /// // for events
+    /// match callsite_enabled() {
+    ///     ALWAYS => (),
+    ///     SOMETIMES => if !enabled() { on_event() },
+    ///     NEVER => on_event(),
+    /// }
+    /// ```
+    ///
+    /// A proper inversion would do `!(enabled() && event_enabled())` (or
+    /// `!enabled() || !event_enabled()`), but because of the implicit `&&`
+    /// relation between `enabled` and `event_enabled`, it is difficult to
+    /// short circuit and not call the wrapped `event_enabled`.
+    ///
+    /// A combinator which remembers the result of `enabled` in order to call
+    /// `event_enabled` only when `enabled() == true` is possible, but requires
+    /// additional thread-local mutable state to support a very niche use case.
+    //
+    //  Also, it'd mean the wrapped layer's `enabled()` always gets called and
+    //  globally applied to events where it doesn't today, since we can't know
+    //  what `event_enabled` will say until we have the event to call it with.
+    ///
+    /// [`Filter`]: crate::layer::Filter
+    /// [`enabled`]: crate::layer::Filter::enabled
+    /// [`event_enabled`]: crate::layer::Filter::event_enabled
+    /// [`callsite_enabled`]: crate::layer::Filter::callsite_enabled
+    pub(crate) fn new(a: A) -> Self {
+        Self { a, _s: PhantomData }
+    }
+}
+
+impl<A, S> Filter<S> for Not<A, S>
+where
+    A: Filter<S>,
+{
+    #[inline]
+    fn enabled(&self, meta: &Metadata<'_>, cx: &Context<'_, S>) -> bool {
+        !self.a.enabled(meta, cx)
+    }
+
+    fn callsite_enabled(&self, meta: &'static Metadata<'static>) -> Interest {
+        match self.a.callsite_enabled(meta) {
+            i if i.is_always() => Interest::never(),
+            i if i.is_never() => Interest::always(),
+            _ => Interest::sometimes(),
+        }
+    }
+
+    fn max_level_hint(&self) -> Option<LevelFilter> {
+        // TODO(eliza): figure this out???
+        None
+    }
+
+    #[inline]
+    fn event_enabled(&self, event: &tracing_core::Event<'_>, cx: &Context<'_, S>) -> bool {
+        // Never disable based on event_enabled; we "disabled" it in `enabled`,
+        // so the `not` has already been applied and filtered this not out.
+        let _ = (event, cx);
+        true
+    }
+
+    #[inline]
+    fn on_new_span(&self, attrs: &Attributes<'_>, id: &Id, ctx: Context<'_, S>) {
+        self.a.on_new_span(attrs, id, ctx);
+    }
+
+    #[inline]
+    fn on_record(&self, id: &Id, values: &Record<'_>, ctx: Context<'_, S>) {
+        self.a.on_record(id, values, ctx.clone());
+    }
+
+    #[inline]
+    fn on_enter(&self, id: &Id, ctx: Context<'_, S>) {
+        self.a.on_enter(id, ctx);
+    }
+
+    #[inline]
+    fn on_exit(&self, id: &Id, ctx: Context<'_, S>) {
+        self.a.on_exit(id, ctx);
+    }
+
+    #[inline]
+    fn on_close(&self, id: Id, ctx: Context<'_, S>) {
+        self.a.on_close(id, ctx);
+    }
+}
+
+impl<A, S> Clone for Not<A, S>
+where
+    A: Clone,
+{
+    fn clone(&self) -> Self {
+        Self {
+            a: self.a.clone(),
+            _s: PhantomData,
+        }
+    }
+}
+
+impl<A, S> fmt::Debug for Not<A, S>
+where
+    A: fmt::Debug,
+{
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        f.debug_tuple("Not").field(&self.a).finish()
+    }
+}

vendor/tracing-subscriber/src/filter/level.rs

@@ -0,0 +1,27 @@
+use tracing_core::{
+    subscriber::{Interest, Subscriber},
+    Metadata,
+};
+
+#[allow(unreachable_pub)] // https://github.com/rust-lang/rust/issues/57411
+pub use tracing_core::metadata::{LevelFilter, ParseLevelFilterError as ParseError};
+
+// === impl LevelFilter ===
+
+impl<S: Subscriber> crate::Layer<S> for LevelFilter {
+    fn register_callsite(&self, metadata: &'static Metadata<'static>) -> Interest {
+        if self >= metadata.level() {
+            Interest::always()
+        } else {
+            Interest::never()
+        }
+    }
+
+    fn enabled(&self, metadata: &Metadata<'_>, _: crate::layer::Context<'_, S>) -> bool {
+        self >= metadata.level()
+    }
+
+    fn max_level_hint(&self) -> Option<LevelFilter> {
+        Some(*self)
+    }
+}

vendor/tracing-subscriber/src/filter/mod.rs

@@ -0,0 +1,66 @@
+//! [`Layer`]s that control which spans and events are enabled by the wrapped
+//! subscriber.
+//!
+//! This module contains a number of types that provide implementations of
+//! various strategies for filtering which spans and events are enabled. For
+//! details on filtering spans and events using [`Layer`]s, see the
+//! [`layer` module's documentation].
+//!
+//! [`layer` module's documentation]: crate::layer#filtering-with-layers
+//! [`Layer`]: crate::layer
+mod filter_fn;
+
+feature! {
+    #![all(feature = "env-filter", feature = "std")]
+    mod env;
+    pub use self::env::*;
+}
+
+feature! {
+    #![all(feature = "registry", feature = "std")]
+    mod layer_filters;
+    pub use self::layer_filters::*;
+}
+
+mod level;
+
+pub use self::filter_fn::*;
+pub use self::level::{LevelFilter, ParseError as LevelParseError};
+
+#[cfg(not(all(feature = "registry", feature = "std")))]
+pub(crate) use self::has_plf_stubs::*;
+
+feature! {
+    #![any(feature = "std", feature = "alloc")]
+    pub mod targets;
+    pub use self::targets::Targets;
+
+    mod directive;
+    pub use self::directive::ParseError;
+}
+
+/// Stub implementations of the per-layer-filter detection functions for when the
+/// `registry` feature is disabled.
+#[cfg(not(all(feature = "registry", feature = "std")))]
+mod has_plf_stubs {
+    pub(crate) fn is_plf_downcast_marker(_: core::any::TypeId) -> bool {
+        false
+    }
+
+    /// Does a type implementing `Subscriber` contain any per-layer filters?
+    pub(crate) fn subscriber_has_plf<S>(_: &S) -> bool
+    where
+        S: tracing_core::Subscriber,
+    {
+        false
+    }
+
+    /// Does a type implementing `Layer` contain any per-layer filters?
+    pub(crate) fn layer_has_plf<L, S>(_: &L) -> bool
+    where
+        L: crate::Layer<S>,
+        S: tracing_core::Subscriber,
+    {
+        false
+    }
+}

vendor/tracing-subscriber/src/filter/targets.rs

@@ -0,0 +1,834 @@
+//! A [filter] that enables or disables spans and events based on their [target] and [level].
+//!
+//! See [`Targets`] for details.
+//!
+//! [target]: tracing_core::Metadata::target
+//! [level]: tracing_core::Level
+//! [filter]: crate::layer#filtering-with-layers
+
+use crate::{
+    filter::{
+        directive::{DirectiveSet, ParseError, StaticDirective},
+        LevelFilter,
+    },
+    layer,
+};
+#[cfg(not(feature = "std"))]
+use alloc::string::String;
+use core::{
+    fmt,
+    iter::{Extend, FilterMap, FromIterator},
+    slice,
+    str::FromStr,
+};
+use tracing_core::{Interest, Level, Metadata, Subscriber};
+
+/// A filter that enables or disables spans and events based on their [target]
+/// and [level].
+///
+/// Targets are typically equal to the Rust module path of the code where the
+/// span or event was recorded, although they may be overridden.
+///
+/// This type can be used for both [per-layer filtering][plf] (using its
+/// [`Filter`] implementation) and [global filtering][global] (using its
+/// [`Layer`] implementation).
+///
+/// See the [documentation on filtering with layers][filtering] for details.
+///
+/// # Filtering With `Targets`
+///
+/// A `Targets` filter consists of one or more [target] prefixes, paired with
+/// [`LevelFilter`]s. If a span or event's [target] begins with one of those
+/// prefixes, and its [level] is at or below the [`LevelFilter`] enabled for
+/// that prefix, then the span or event will be enabled.
+///
+/// This is similar to the behavior implemented by the [`env_logger` crate] in
+/// the `log` ecosystem.
+///
+/// The [`EnvFilter`] type also provided by this crate is very similar to `Targets`,
+/// but is capable of a more sophisticated form of filtering where events may
+/// also be enabled or disabled based on the span they are recorded in.
+/// `Targets` can be thought of as a lighter-weight form of [`EnvFilter`] that
+/// can be used instead when this dynamic filtering is not required.
+///
+/// # Examples
+///
+/// A `Targets` filter can be constructed by programmatically adding targets and
+/// levels to enable:
+///
+/// ```
+/// use tracing_subscriber::{filter, prelude::*};
+/// use tracing_core::Level;
+///
+/// let filter = filter::Targets::new()
+///     // Enable the `INFO` level for anything in `my_crate`
+///     .with_target("my_crate", Level::INFO)
+///     // Enable the `DEBUG` level for a specific module.
+///     .with_target("my_crate::interesting_module", Level::DEBUG);
+///
+/// // Build a new subscriber with the `fmt` layer using the `Targets`
+/// // filter we constructed above.
+/// tracing_subscriber::registry()
+///     .with(tracing_subscriber::fmt::layer())
+///     .with(filter)
+///     .init();
+/// ```
+///
+/// [`LevelFilter::OFF`] can be used to disable a particular target:
+/// ```
+/// use tracing_subscriber::filter::{Targets, LevelFilter};
+/// use tracing_core::Level;
+///
+/// let filter = Targets::new()
+///     .with_target("my_crate", Level::INFO)
+///     // Disable all traces from `annoying_module`.
+///     .with_target("my_crate::annoying_module", LevelFilter::OFF);
+/// # drop(filter);
+/// ```
+///
+/// Alternatively, `Targets` implements [`std::str::FromStr`], allowing it to be
+/// parsed from a comma-delimited list of `target=level` pairs. For example:
+///
+/// ```rust
+/// # fn main() -> Result<(), Box<dyn std::error::Error>> {
+/// use tracing_subscriber::filter;
+/// use tracing_core::Level;
+///
+/// let filter = "my_crate=info,my_crate::interesting_module=trace,other_crate=debug"
+///     .parse::<filter::Targets>()?;
+///
+/// // The parsed filter is identical to a filter constructed using `with_target`:
+/// assert_eq!(
+///     filter,
+///     filter::Targets::new()
+///         .with_target("my_crate", Level::INFO)
+///         .with_target("my_crate::interesting_module", Level::TRACE)
+///         .with_target("other_crate", Level::DEBUG)
+/// );
+/// # Ok(()) }
+/// ```
+///
+/// This is particularly useful when the list of enabled targets is configurable
+/// by the user at runtime.
+///
+/// The `Targets` filter can be used as a [per-layer filter][plf] *and* as a
+/// [global filter][global]:
+///
+/// ```rust
+/// use tracing_subscriber::{
+///     fmt,
+///     filter::{Targets, LevelFilter},
+///     prelude::*,
+/// };
+/// use tracing_core::Level;
+/// use std::{sync::Arc, fs::File};
+/// # fn docs() -> Result<(), Box<dyn std::error::Error>> {
+///
+/// // A layer that logs events to stdout using the human-readable "pretty"
+/// // format.
+/// let stdout_log = fmt::layer().pretty();
+///
+/// // A layer that logs events to a file, using the JSON format.
+/// let file = File::create("debug_log.json")?;
+/// let debug_log = fmt::layer()
+///     .with_writer(Arc::new(file))
+///     .json();
+///
+/// tracing_subscriber::registry()
+///     // Only log INFO and above to stdout, unless the span or event
+///     // has the `my_crate::cool_module` target prefix.
+///     .with(stdout_log
+///         .with_filter(
+///             Targets::default()
+///                 .with_target("my_crate::cool_module", Level::DEBUG)
+///                 .with_default(Level::INFO)
+///        )
+///     )
+///     // Log everything enabled by the global filter to `debug_log.json`.
+///     .with(debug_log)
+///     // Configure a global filter for the whole subscriber stack. This will
+///     // control what spans and events are recorded by both the `debug_log`
+///     // and the `stdout_log` layers, and `stdout_log` will *additionally* be
+///     // filtered by its per-layer filter.
+///     .with(
+///         Targets::default()
+///             .with_target("my_crate", Level::TRACE)
+///             .with_target("other_crate", Level::INFO)
+///             .with_target("other_crate::annoying_module", LevelFilter::OFF)
+///             .with_target("third_crate", Level::DEBUG)
+///     ).init();
+/// # Ok(()) }
+///```
+///
+/// [target]: tracing_core::Metadata::target
+/// [level]: tracing_core::Level
+/// [`Filter`]: crate::layer::Filter
+/// [`Layer`]: crate::layer::Layer
+/// [plf]: crate::layer#per-layer-filtering
+/// [global]: crate::layer#global-filtering
+/// [filtering]: crate::layer#filtering-with-layers
+/// [`env_logger` crate]: https://docs.rs/env_logger/0.9.0/env_logger/index.html#enabling-logging
+/// [`EnvFilter`]: crate::filter::EnvFilter
+#[derive(Debug, Default, Clone, PartialEq)]
+pub struct Targets(DirectiveSet<StaticDirective>);
+
+impl Targets {
+    /// Returns a new `Targets` filter.
+    ///
+    /// This filter will enable no targets. Call [`with_target`] or [`with_targets`]
+    /// to add enabled targets, and [`with_default`] to change the default level
+    /// enabled for spans and events that didn't match any of the provided targets.
+    ///
+    /// [`with_target`]: Targets::with_target
+    /// [`with_targets`]: Targets::with_targets
+    /// [`with_default`]: Targets::with_default
+    pub fn new() -> Self {
+        Self::default()
+    }
+
+    /// Enables spans and events with [target]s starting with the provided target
+    /// prefix if they are at or below the provided [`LevelFilter`].
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// use tracing_subscriber::filter;
+    /// use tracing_core::Level;
+    ///
+    /// let filter = filter::Targets::new()
+    ///     // Enable the `INFO` level for anything in `my_crate`
+    ///     .with_target("my_crate", Level::INFO)
+    ///     // Enable the `DEBUG` level for a specific module.
+    ///     .with_target("my_crate::interesting_module", Level::DEBUG);
+    /// # drop(filter);
+    /// ```
+    ///
+    /// [`LevelFilter::OFF`] can be used to disable a particular target:
+    /// ```
+    /// use tracing_subscriber::filter::{Targets, LevelFilter};
+    /// use tracing_core::Level;
+    ///
+    /// let filter = Targets::new()
+    ///     .with_target("my_crate", Level::INFO)
+    ///     // Disable all traces from `annoying_module`.
+    ///     .with_target("my_crate::interesting_module", LevelFilter::OFF);
+    /// # drop(filter);
+    /// ```
+    ///
+    /// [target]: tracing_core::Metadata::target
+    pub fn with_target(mut self, target: impl Into<String>, level: impl Into<LevelFilter>) -> Self {
+        self.0.add(StaticDirective::new(
+            Some(target.into()),
+            Default::default(),
+            level.into(),
+        ));
+        self
+    }
+    /// Adds [target]s from an iterator of [target]-[`LevelFilter`] pairs to this filter.
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// use tracing_subscriber::filter;
+    /// use tracing_core::Level;
+    ///
+    /// let filter = filter::Targets::new()
+    ///     .with_targets(vec![
+    ///         ("my_crate", Level::INFO),
+    ///         ("my_crate::some_module", Level::DEBUG),
+    ///         ("my_crate::other_module::cool_stuff", Level::TRACE),
+    ///         ("other_crate", Level::WARN)
+    ///     ]);
+    /// # drop(filter);
+    /// ```
+    ///
+    /// [`LevelFilter::OFF`] can be used to disable a particular target:
+    /// ```
+    /// use tracing_subscriber::filter::{Targets, LevelFilter};
+    /// use tracing_core::Level;
+    ///
+    /// let filter = Targets::new()
+    ///     .with_target("my_crate", Level::INFO)
+    ///     // Disable all traces from `annoying_module`.
+    ///     .with_target("my_crate::interesting_module", LevelFilter::OFF);
+    /// # drop(filter);
+    /// ```
+    ///
+    /// [target]: tracing_core::Metadata::target
+    pub fn with_targets<T, L>(mut self, targets: impl IntoIterator<Item = (T, L)>) -> Self
+    where
+        String: From<T>,
+        LevelFilter: From<L>,
+    {
+        self.extend(targets);
+        self
+    }
+
+    /// Sets the default level to enable for spans and events whose targets did
+    /// not match any of the configured prefixes.
+    ///
+    /// By default, this is [`LevelFilter::OFF`]. This means that spans and
+    /// events will only be enabled if they match one of the configured target
+    /// prefixes. If this is changed to a different [`LevelFilter`], spans and
+    /// events with targets that did not match any of the configured prefixes
+    /// will be enabled if their level is at or below the provided level.
+    pub fn with_default(mut self, level: impl Into<LevelFilter>) -> Self {
+        self.0
+            .add(StaticDirective::new(None, Default::default(), level.into()));
+        self
+    }
+
+    /// Returns the default level for this filter, if one is set.
+    ///
+    /// The default level is used to filter any spans or events with targets
+    /// that do not match any of the configured set of prefixes.
+    ///
+    /// The default level can be set for a filter either by using
+    /// [`with_default`](Self::with_default) or when parsing from a filter string that includes a
+    /// level without a target (e.g. `"trace"`).
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// use tracing_subscriber::filter::{LevelFilter, Targets};
+    ///
+    /// let filter = Targets::new().with_default(LevelFilter::INFO);
+    /// assert_eq!(filter.default_level(), Some(LevelFilter::INFO));
+    ///
+    /// let filter: Targets = "info".parse().unwrap();
+    /// assert_eq!(filter.default_level(), Some(LevelFilter::INFO));
+    /// ```
+    ///
+    /// The default level is `None` if no default is set:
+    ///
+    /// ```
+    /// use tracing_subscriber::filter::Targets;
+    ///
+    /// let filter = Targets::new();
+    /// assert_eq!(filter.default_level(), None);
+    ///
+    /// let filter: Targets = "my_crate=info".parse().unwrap();
+    /// assert_eq!(filter.default_level(), None);
+    /// ```
+    ///
+    /// Note that an unset default level (`None`) behaves like [`LevelFilter::OFF`] when the filter is
+    /// used, but it could also be set explicitly which may be useful to distinguish (such as when
+    /// merging multiple `Targets`).
+    ///
+    /// ```
+    /// use tracing_subscriber::filter::{LevelFilter, Targets};
+    ///
+    /// let filter = Targets::new().with_default(LevelFilter::OFF);
+    /// assert_eq!(filter.default_level(), Some(LevelFilter::OFF));
+    ///
+    /// let filter: Targets = "off".parse().unwrap();
+    /// assert_eq!(filter.default_level(), Some(LevelFilter::OFF));
+    /// ```
+    pub fn default_level(&self) -> Option<LevelFilter> {
+        self.0.directives().find_map(|d| {
+            if d.target.is_none() {
+                Some(d.level)
+            } else {
+                None
+            }
+        })
+    }
+
+    /// Returns an iterator over the [target]-[`LevelFilter`] pairs in this filter.
+    ///
+    /// The order of iteration is undefined.
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// use tracing_subscriber::filter::{Targets, LevelFilter};
+    /// use tracing_core::Level;
+    ///
+    /// let filter = Targets::new()
+    ///     .with_target("my_crate", Level::INFO)
+    ///     .with_target("my_crate::interesting_module", Level::DEBUG);
+    ///
+    /// let mut targets: Vec<_> = filter.iter().collect();
+    /// targets.sort();
+    ///
+    /// assert_eq!(targets, vec![
+    ///     ("my_crate", LevelFilter::INFO),
+    ///     ("my_crate::interesting_module", LevelFilter::DEBUG),
+    /// ]);
+    /// ```
+    ///
+    /// [target]: tracing_core::Metadata::target
+    pub fn iter(&self) -> Iter<'_> {
+        self.into_iter()
+    }
+
+    #[inline]
+    fn interested(&self, metadata: &'static Metadata<'static>) -> Interest {
+        if self.0.enabled(metadata) {
+            Interest::always()
+        } else {
+            Interest::never()
+        }
+    }
+
+    /// Returns whether a [target]-[`Level`] pair would be enabled
+    /// by this `Targets`.
+    ///
+    /// This method can be used with [`module_path!`] from `std` as the target
+    /// in order to emulate the behavior of the [`tracing::event!`] and [`tracing::span!`]
+    /// macros.
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// use tracing_subscriber::filter::{Targets, LevelFilter};
+    /// use tracing_core::Level;
+    ///
+    /// let filter = Targets::new()
+    ///     .with_target("my_crate", Level::INFO)
+    ///     .with_target("my_crate::interesting_module", Level::DEBUG);
+    ///
+    /// assert!(filter.would_enable("my_crate", &Level::INFO));
+    /// assert!(!filter.would_enable("my_crate::interesting_module", &Level::TRACE));
+    /// ```
+    ///
+    /// [target]: tracing_core::Metadata::target
+    /// [`module_path!`]: std::module_path!
+    pub fn would_enable(&self, target: &str, level: &Level) -> bool {
+        // "Correct" to call because `Targets` only produces `StaticDirective`'s with NO
+        // fields
+        self.0.target_enabled(target, level)
+    }
+}
+
+impl<T, L> Extend<(T, L)> for Targets
+where
+    T: Into<String>,
+    L: Into<LevelFilter>,
+{
+    fn extend<I: IntoIterator<Item = (T, L)>>(&mut self, iter: I) {
+        let iter = iter.into_iter().map(|(target, level)| {
+            StaticDirective::new(Some(target.into()), Default::default(), level.into())
+        });
+        self.0.extend(iter);
+    }
+}
+
+impl<T, L> FromIterator<(T, L)> for Targets
+where
+    T: Into<String>,
+    L: Into<LevelFilter>,
+{
+    fn from_iter<I: IntoIterator<Item = (T, L)>>(iter: I) -> Self {
+        let mut this = Self::default();
+        this.extend(iter);
+        this
+    }
+}
+
+impl FromStr for Targets {
+    type Err = ParseError;
+    fn from_str(s: &str) -> Result<Self, Self::Err> {
+        s.split(',')
+            .map(StaticDirective::from_str)
+            .collect::<Result<_, _>>()
+            .map(Self)
+    }
+}
+
+impl<S> layer::Layer<S> for Targets
+where
+    S: Subscriber,
+{
+    fn enabled(&self, metadata: &Metadata<'_>, _: layer::Context<'_, S>) -> bool {
+        self.0.enabled(metadata)
+    }
+
+    fn register_callsite(&self, metadata: &'static Metadata<'static>) -> Interest {
+        self.interested(metadata)
+    }
+
+    fn max_level_hint(&self) -> Option<LevelFilter> {
+        Some(self.0.max_level)
+    }
+}
+
+#[cfg(feature = "registry")]
+#[cfg_attr(docsrs, doc(cfg(feature = "registry")))]
+impl<S> layer::Filter<S> for Targets {
+    fn enabled(&self, metadata: &Metadata<'_>, _: &layer::Context<'_, S>) -> bool {
+        self.0.enabled(metadata)
+    }
+
+    fn callsite_enabled(&self, metadata: &'static Metadata<'static>) -> Interest {
+        self.interested(metadata)
+    }
+
+    fn max_level_hint(&self) -> Option<LevelFilter> {
+        Some(self.0.max_level)
+    }
+}
+
+impl IntoIterator for Targets {
+    type Item = (String, LevelFilter);
+
+    type IntoIter = IntoIter;
+
+    fn into_iter(self) -> Self::IntoIter {
+        IntoIter::new(self)
+    }
+}
+
+impl<'a> IntoIterator for &'a Targets {
+    type Item = (&'a str, LevelFilter);
+
+    type IntoIter = Iter<'a>;
+
+    fn into_iter(self) -> Self::IntoIter {
+        Iter::new(self)
+    }
+}
+
+impl fmt::Display for Targets {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        let mut directives = self.0.directives();
+        if let Some(directive) = directives.next() {
+            write!(f, "{}", directive)?;
+            for directive in directives {
+                write!(f, ",{}", directive)?;
+            }
+        }
+
+        Ok(())
+    }
+}
+
+/// An owning iterator over the [target]-[level] pairs of a `Targets` filter.
+///
+/// This struct is created by the `IntoIterator` trait implementation of [`Targets`].
+///
+/// # Examples
+///
+/// Merge the targets from one `Targets` with another:
+///
+/// ```
+/// use tracing_subscriber::filter::Targets;
+/// use tracing_core::Level;
+///
+/// let mut filter = Targets::new().with_target("my_crate", Level::INFO);
+/// let overrides = Targets::new().with_target("my_crate::interesting_module", Level::DEBUG);
+///
+/// filter.extend(overrides);
+/// # drop(filter);
+/// ```
+///
+/// [target]: tracing_core::Metadata::target
+/// [level]: tracing_core::Level
+#[derive(Debug)]
+pub struct IntoIter(
+    #[allow(clippy::type_complexity)] // alias indirection would probably make this more confusing
+    FilterMap<
+        <DirectiveSet<StaticDirective> as IntoIterator>::IntoIter,
+        fn(StaticDirective) -> Option<(String, LevelFilter)>,
+    >,
+);
+
+impl IntoIter {
+    fn new(targets: Targets) -> Self {
+        Self(targets.0.into_iter().filter_map(|directive| {
+            let level = directive.level;
+            directive.target.map(|target| (target, level))
+        }))
+    }
+}
+
+impl Iterator for IntoIter {
+    type Item = (String, LevelFilter);
+
+    fn next(&mut self) -> Option<Self::Item> {
+        self.0.next()
+    }
+
+    fn size_hint(&self) -> (usize, Option<usize>) {
+        self.0.size_hint()
+    }
+}
+
+/// A borrowing iterator over the [target]-[level] pairs of a `Targets` filter.
+///
+/// This struct is created by [`iter`] method of [`Targets`], or from the `IntoIterator`
+/// implementation for `&Targets`.
+///
+/// [target]: tracing_core::Metadata::target
+/// [level]: tracing_core::Level
+/// [`iter`]: Targets::iter
+#[derive(Debug)]
+pub struct Iter<'a>(
+    FilterMap<
+        slice::Iter<'a, StaticDirective>,
+        fn(&'a StaticDirective) -> Option<(&'a str, LevelFilter)>,
+    >,
+);
+
+impl<'a> Iter<'a> {
+    fn new(targets: &'a Targets) -> Self {
+        Self(targets.0.iter().filter_map(|directive| {
+            directive
+                .target
+                .as_deref()
+                .map(|target| (target, directive.level))
+        }))
+    }
+}
+
+impl<'a> Iterator for Iter<'a> {
+    type Item = (&'a str, LevelFilter);
+
+    fn next(&mut self) -> Option<Self::Item> {
+        self.0.next()
+    }
+
+    fn size_hint(&self) -> (usize, Option<usize>) {
+        self.0.size_hint()
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    feature! {
+        #![not(feature = "std")]
+        use alloc::{vec, vec::Vec, string::ToString};
+
+        // `dbg!` is only available with `libstd`; just nop it out when testing
+        // with alloc only.
+        macro_rules! dbg {
+            ($x:expr) => { $x }
+        }
+    }
+
+    fn expect_parse(s: &str) -> Targets {
+        match dbg!(s).parse::<Targets>() {
+            Err(e) => panic!("string {:?} did not parse successfully: {}", s, e),
+            Ok(e) => e,
+        }
+    }
+
+    fn expect_parse_ralith(s: &str) {
+        let dirs = expect_parse(s).0.into_vec();
+        assert_eq!(dirs.len(), 2, "\nparsed: {:#?}", dirs);
+        assert_eq!(dirs[0].target, Some("server".to_string()));
+        assert_eq!(dirs[0].level, LevelFilter::DEBUG);
+        assert_eq!(dirs[0].field_names, Vec::<String>::new());
+
+        assert_eq!(dirs[1].target, Some("common".to_string()));
+        assert_eq!(dirs[1].level, LevelFilter::INFO);
+        assert_eq!(dirs[1].field_names, Vec::<String>::new());
+    }
+
+    fn expect_parse_level_directives(s: &str) {
+        let dirs = expect_parse(s).0.into_vec();
+        assert_eq!(dirs.len(), 6, "\nparsed: {:#?}", dirs);
+
+        assert_eq!(dirs[0].target, Some("crate3::mod2::mod1".to_string()));
+        assert_eq!(dirs[0].level, LevelFilter::OFF);
+        assert_eq!(dirs[0].field_names, Vec::<String>::new());
+
+        assert_eq!(dirs[1].target, Some("crate1::mod2::mod3".to_string()));
+        assert_eq!(dirs[1].level, LevelFilter::INFO);
+        assert_eq!(dirs[1].field_names, Vec::<String>::new());
+
+        assert_eq!(dirs[2].target, Some("crate1::mod2".to_string()));
+        assert_eq!(dirs[2].level, LevelFilter::WARN);
+        assert_eq!(dirs[2].field_names, Vec::<String>::new());
+
+        assert_eq!(dirs[3].target, Some("crate1::mod1".to_string()));
+        assert_eq!(dirs[3].level, LevelFilter::ERROR);
+        assert_eq!(dirs[3].field_names, Vec::<String>::new());
+
+        assert_eq!(dirs[4].target, Some("crate3".to_string()));
+        assert_eq!(dirs[4].level, LevelFilter::TRACE);
+        assert_eq!(dirs[4].field_names, Vec::<String>::new());
+
+        assert_eq!(dirs[5].target, Some("crate2".to_string()));
+        assert_eq!(dirs[5].level, LevelFilter::DEBUG);
+        assert_eq!(dirs[5].field_names, Vec::<String>::new());
+    }
+
+    #[test]
+    fn parse_ralith() {
+        expect_parse_ralith("common=info,server=debug");
+    }
+
+    #[test]
+    fn parse_ralith_uc() {
+        expect_parse_ralith("common=INFO,server=DEBUG");
+    }
+
+    #[test]
+    fn parse_ralith_mixed() {
+        expect_parse("common=iNfo,server=dEbUg");
+    }
+
+    #[test]
+    fn expect_parse_valid() {
+        let dirs = expect_parse("crate1::mod1=error,crate1::mod2,crate2=debug,crate3=off")
+            .0
+            .into_vec();
+        assert_eq!(dirs.len(), 4, "\nparsed: {:#?}", dirs);
+        assert_eq!(dirs[0].target, Some("crate1::mod2".to_string()));
+        assert_eq!(dirs[0].level, LevelFilter::TRACE);
+        assert_eq!(dirs[0].field_names, Vec::<String>::new());
+
+        assert_eq!(dirs[1].target, Some("crate1::mod1".to_string()));
+        assert_eq!(dirs[1].level, LevelFilter::ERROR);
+        assert_eq!(dirs[1].field_names, Vec::<String>::new());
+
+        assert_eq!(dirs[2].target, Some("crate3".to_string()));
+        assert_eq!(dirs[2].level, LevelFilter::OFF);
+        assert_eq!(dirs[2].field_names, Vec::<String>::new());
+
+        assert_eq!(dirs[3].target, Some("crate2".to_string()));
+        assert_eq!(dirs[3].level, LevelFilter::DEBUG);
+        assert_eq!(dirs[3].field_names, Vec::<String>::new());
+    }
+
+    #[test]
+    fn parse_level_directives() {
+        expect_parse_level_directives(
+            "crate1::mod1=error,crate1::mod2=warn,crate1::mod2::mod3=info,\
+             crate2=debug,crate3=trace,crate3::mod2::mod1=off",
+        )
+    }
+
+    #[test]
+    fn parse_uppercase_level_directives() {
+        expect_parse_level_directives(
+            "crate1::mod1=ERROR,crate1::mod2=WARN,crate1::mod2::mod3=INFO,\
+             crate2=DEBUG,crate3=TRACE,crate3::mod2::mod1=OFF",
+        )
+    }
+
+    #[test]
+    fn parse_numeric_level_directives() {
+        expect_parse_level_directives(
+            "crate1::mod1=1,crate1::mod2=2,crate1::mod2::mod3=3,crate2=4,\
+             crate3=5,crate3::mod2::mod1=0",
+        )
+    }
+
+    #[test]
+    fn targets_iter() {
+        let filter = expect_parse("crate1::mod1=error,crate1::mod2,crate2=debug,crate3=off")
+            .with_default(LevelFilter::WARN);
+
+        let mut targets: Vec<_> = filter.iter().collect();
+        targets.sort();
+
+        assert_eq!(
+            targets,
+            vec![
+                ("crate1::mod1", LevelFilter::ERROR),
+                ("crate1::mod2", LevelFilter::TRACE),
+                ("crate2", LevelFilter::DEBUG),
+                ("crate3", LevelFilter::OFF),
+            ]
+        );
+    }
+
+    #[test]
+    fn targets_into_iter() {
+        let filter = expect_parse("crate1::mod1=error,crate1::mod2,crate2=debug,crate3=off")
+            .with_default(LevelFilter::WARN);
+
+        let mut targets: Vec<_> = filter.into_iter().collect();
+        targets.sort();
+
+        assert_eq!(
+            targets,
+            vec![
+                ("crate1::mod1".to_string(), LevelFilter::ERROR),
+                ("crate1::mod2".to_string(), LevelFilter::TRACE),
+                ("crate2".to_string(), LevelFilter::DEBUG),
+                ("crate3".to_string(), LevelFilter::OFF),
+            ]
+        );
+    }
+
+    #[test]
+    fn targets_default_level() {
+        let filter = expect_parse("crate1::mod1=error,crate1::mod2,crate2=debug,crate3=off");
+        assert_eq!(filter.default_level(), None);
+
+        let filter = expect_parse("crate1::mod1=error,crate1::mod2,crate2=debug,crate3=off")
+            .with_default(LevelFilter::OFF);
+        assert_eq!(filter.default_level(), Some(LevelFilter::OFF));
+
+        let filter = expect_parse("crate1::mod1=error,crate1::mod2,crate2=debug,crate3=off")
+            .with_default(LevelFilter::OFF)
+            .with_default(LevelFilter::INFO);
+        assert_eq!(filter.default_level(), Some(LevelFilter::INFO));
+    }
+
+    #[test]
+    // `println!` is only available with `libstd`.
+    #[cfg(feature = "std")]
+    fn size_of_filters() {
+        fn print_sz(s: &str) {
+            let filter = s.parse::<Targets>().expect("filter should parse");
+            println!(
+                "size_of_val({:?})\n -> {}B",
+                s,
+                std::mem::size_of_val(&filter)
+            );
+        }
+
+        print_sz("info");
+
+        print_sz("foo=debug");
+
+        print_sz(
+            "crate1::mod1=error,crate1::mod2=warn,crate1::mod2::mod3=info,\
+            crate2=debug,crate3=trace,crate3::mod2::mod1=off",
+        );
+    }
+
+    /// Test that the `fmt::Display` implementation for `Targets` emits a string
+    /// that can itself be parsed as a `Targets`, and that the parsed `Targets`
+    /// is equivalent to the original one.
+    #[test]
+    fn display_roundtrips() {
+        fn test_roundtrip(s: &str) {
+            let filter = expect_parse(s);
+            // we don't assert that the display output is equivalent to the
+            // original parsed filter string, because the `Display` impl always
+            // uses lowercase level names and doesn't use the
+            // target-without-level shorthand syntax. while they may not be
+            // textually equivalent, though, they should still *parse* to the
+            // same filter.
+            let formatted = filter.to_string();
+            let filter2 = match dbg!(&formatted).parse::<Targets>() {
+                Ok(filter) => filter,
+                Err(e) => panic!(
+                    "failed to parse formatted filter string {:?}: {}",
+                    formatted, e
+                ),
+            };
+            assert_eq!(filter, filter2);
+        }
+
+        test_roundtrip("crate1::mod1=error,crate1::mod2,crate2=debug,crate3=off");
+        test_roundtrip(
+            "crate1::mod1=ERROR,crate1::mod2=WARN,crate1::mod2::mod3=INFO,\
+        crate2=DEBUG,crate3=TRACE,crate3::mod2::mod1=OFF",
+        );
+        test_roundtrip(
+            "crate1::mod1=error,crate1::mod2=warn,crate1::mod2::mod3=info,\
+             crate2=debug,crate3=trace,crate3::mod2::mod1=off",
+        );
+        test_roundtrip("crate1::mod1,crate1::mod2,info");
+        test_roundtrip("crate1");
+        test_roundtrip("info");
+    }
+}

vendor/tracing-subscriber/src/fmt/format/escape.rs

@@ -0,0 +1,51 @@
+//! ANSI escape sequence sanitization to prevent terminal injection attacks.
+
+use std::fmt::{self, Write};
+
+/// A wrapper that implements `fmt::Debug` and `fmt::Display` and escapes ANSI sequences on-the-fly.
+/// This avoids creating intermediate strings while providing security against terminal injection.
+pub(super) struct Escape<T>(pub(super) T);
+
+/// Helper struct that escapes ANSI sequences as characters are written
+struct EscapingWriter<'a, 'b> {
+    inner: &'a mut fmt::Formatter<'b>,
+}
+
+impl<'a, 'b> fmt::Write for EscapingWriter<'a, 'b> {
+    fn write_str(&mut self, s: &str) -> fmt::Result {
+        // Stream the string character by character, escaping ANSI and C1 control sequences
+        for ch in s.chars() {
+            match ch {
+                // C0 control characters that can be used in terminal escape sequences
+                '\x1b' => self.inner.write_str("\\x1b")?,  // ESC
+                '\x07' => self.inner.write_str("\\x07")?,  // BEL
+                '\x08' => self.inner.write_str("\\x08")?,  // BS
+                '\x0c' => self.inner.write_str("\\x0c")?,  // FF
+                '\x7f' => self.inner.write_str("\\x7f")?,  // DEL
+                
+                // C1 control characters (\x80-\x9f) - 8-bit control codes
+                // These can be used as alternative escape sequences in some terminals
+                ch if ch as u32 >= 0x80 && ch as u32 <= 0x9f => {
+                    write!(self.inner, "\\u{{{:x}}}", ch as u32)?
+                },
+                
+                _ => self.inner.write_char(ch)?,
+            }
+        }
+        Ok(())
+    }
+}
+
+impl<T: fmt::Debug> fmt::Debug for Escape<T> {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        let mut escaping_writer = EscapingWriter { inner: f };
+        write!(escaping_writer, "{:?}", self.0)
+    }
+}
+
+impl<T: fmt::Display> fmt::Display for Escape<T> {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        let mut escaping_writer = EscapingWriter { inner: f };
+        write!(escaping_writer, "{}", self.0)
+    }
+}

vendor/tracing-subscriber/src/fmt/format/json.rs

@@ -0,0 +1,896 @@
+use super::{Format, FormatEvent, FormatFields, FormatTime, Writer};
+use crate::{
+    field::{RecordFields, VisitOutput},
+    fmt::{
+        fmt_layer::{FmtContext, FormattedFields},
+        writer::WriteAdaptor,
+    },
+    registry::LookupSpan,
+};
+use serde::ser::{SerializeMap, Serializer as _};
+use serde_json::Serializer;
+use std::{
+    collections::BTreeMap,
+    fmt::{self, Write},
+};
+use tracing_core::{
+    field::{self, Field},
+    span::Record,
+    Event, Subscriber,
+};
+use tracing_serde::AsSerde;
+
+#[cfg(feature = "tracing-log")]
+use tracing_log::NormalizeEvent;
+
+/// Marker for [`Format`] that indicates that the newline-delimited JSON log
+/// format should be used.
+///
+/// This formatter is intended for production use with systems where structured
+/// logs are consumed as JSON by analysis and viewing tools. The JSON output is
+/// not optimized for human readability; instead, it should be pretty-printed
+/// using external JSON tools such as `jq`, or using a JSON log viewer.
+///
+/// # Example Output
+///
+/// <pre><font color="#4E9A06"><b>:;</b></font> <font color="#4E9A06">cargo</font> run --example fmt-json
+/// <font color="#4E9A06"><b>    Finished</b></font> dev [unoptimized + debuginfo] target(s) in 0.08s
+/// <font color="#4E9A06"><b>     Running</b></font> `target/debug/examples/fmt-json`
+/// {&quot;timestamp&quot;:&quot;2022-02-15T18:47:10.821315Z&quot;,&quot;level&quot;:&quot;INFO&quot;,&quot;fields&quot;:{&quot;message&quot;:&quot;preparing to shave yaks&quot;,&quot;number_of_yaks&quot;:3},&quot;target&quot;:&quot;fmt_json&quot;}
+/// {&quot;timestamp&quot;:&quot;2022-02-15T18:47:10.821422Z&quot;,&quot;level&quot;:&quot;INFO&quot;,&quot;fields&quot;:{&quot;message&quot;:&quot;shaving yaks&quot;},&quot;target&quot;:&quot;fmt_json::yak_shave&quot;,&quot;spans&quot;:[{&quot;yaks&quot;:3,&quot;name&quot;:&quot;shaving_yaks&quot;}]}
+/// {&quot;timestamp&quot;:&quot;2022-02-15T18:47:10.821495Z&quot;,&quot;level&quot;:&quot;TRACE&quot;,&quot;fields&quot;:{&quot;message&quot;:&quot;hello! I&apos;m gonna shave a yak&quot;,&quot;excitement&quot;:&quot;yay!&quot;},&quot;target&quot;:&quot;fmt_json::yak_shave&quot;,&quot;spans&quot;:[{&quot;yaks&quot;:3,&quot;name&quot;:&quot;shaving_yaks&quot;},{&quot;yak&quot;:1,&quot;name&quot;:&quot;shave&quot;}]}
+/// {&quot;timestamp&quot;:&quot;2022-02-15T18:47:10.821546Z&quot;,&quot;level&quot;:&quot;TRACE&quot;,&quot;fields&quot;:{&quot;message&quot;:&quot;yak shaved successfully&quot;},&quot;target&quot;:&quot;fmt_json::yak_shave&quot;,&quot;spans&quot;:[{&quot;yaks&quot;:3,&quot;name&quot;:&quot;shaving_yaks&quot;},{&quot;yak&quot;:1,&quot;name&quot;:&quot;shave&quot;}]}
+/// {&quot;timestamp&quot;:&quot;2022-02-15T18:47:10.821598Z&quot;,&quot;level&quot;:&quot;DEBUG&quot;,&quot;fields&quot;:{&quot;yak&quot;:1,&quot;shaved&quot;:true},&quot;target&quot;:&quot;yak_events&quot;,&quot;spans&quot;:[{&quot;yaks&quot;:3,&quot;name&quot;:&quot;shaving_yaks&quot;}]}
+/// {&quot;timestamp&quot;:&quot;2022-02-15T18:47:10.821637Z&quot;,&quot;level&quot;:&quot;TRACE&quot;,&quot;fields&quot;:{&quot;yaks_shaved&quot;:1},&quot;target&quot;:&quot;fmt_json::yak_shave&quot;,&quot;spans&quot;:[{&quot;yaks&quot;:3,&quot;name&quot;:&quot;shaving_yaks&quot;}]}
+/// {&quot;timestamp&quot;:&quot;2022-02-15T18:47:10.821684Z&quot;,&quot;level&quot;:&quot;TRACE&quot;,&quot;fields&quot;:{&quot;message&quot;:&quot;hello! I&apos;m gonna shave a yak&quot;,&quot;excitement&quot;:&quot;yay!&quot;},&quot;target&quot;:&quot;fmt_json::yak_shave&quot;,&quot;spans&quot;:[{&quot;yaks&quot;:3,&quot;name&quot;:&quot;shaving_yaks&quot;},{&quot;yak&quot;:2,&quot;name&quot;:&quot;shave&quot;}]}
+/// {&quot;timestamp&quot;:&quot;2022-02-15T18:47:10.821727Z&quot;,&quot;level&quot;:&quot;TRACE&quot;,&quot;fields&quot;:{&quot;message&quot;:&quot;yak shaved successfully&quot;},&quot;target&quot;:&quot;fmt_json::yak_shave&quot;,&quot;spans&quot;:[{&quot;yaks&quot;:3,&quot;name&quot;:&quot;shaving_yaks&quot;},{&quot;yak&quot;:2,&quot;name&quot;:&quot;shave&quot;}]}
+/// {&quot;timestamp&quot;:&quot;2022-02-15T18:47:10.821773Z&quot;,&quot;level&quot;:&quot;DEBUG&quot;,&quot;fields&quot;:{&quot;yak&quot;:2,&quot;shaved&quot;:true},&quot;target&quot;:&quot;yak_events&quot;,&quot;spans&quot;:[{&quot;yaks&quot;:3,&quot;name&quot;:&quot;shaving_yaks&quot;}]}
+/// {&quot;timestamp&quot;:&quot;2022-02-15T18:47:10.821806Z&quot;,&quot;level&quot;:&quot;TRACE&quot;,&quot;fields&quot;:{&quot;yaks_shaved&quot;:2},&quot;target&quot;:&quot;fmt_json::yak_shave&quot;,&quot;spans&quot;:[{&quot;yaks&quot;:3,&quot;name&quot;:&quot;shaving_yaks&quot;}]}
+/// {&quot;timestamp&quot;:&quot;2022-02-15T18:47:10.821909Z&quot;,&quot;level&quot;:&quot;TRACE&quot;,&quot;fields&quot;:{&quot;message&quot;:&quot;hello! I&apos;m gonna shave a yak&quot;,&quot;excitement&quot;:&quot;yay!&quot;},&quot;target&quot;:&quot;fmt_json::yak_shave&quot;,&quot;spans&quot;:[{&quot;yaks&quot;:3,&quot;name&quot;:&quot;shaving_yaks&quot;},{&quot;yak&quot;:3,&quot;name&quot;:&quot;shave&quot;}]}
+/// {&quot;timestamp&quot;:&quot;2022-02-15T18:47:10.821956Z&quot;,&quot;level&quot;:&quot;WARN&quot;,&quot;fields&quot;:{&quot;message&quot;:&quot;could not locate yak&quot;},&quot;target&quot;:&quot;fmt_json::yak_shave&quot;,&quot;spans&quot;:[{&quot;yaks&quot;:3,&quot;name&quot;:&quot;shaving_yaks&quot;},{&quot;yak&quot;:3,&quot;name&quot;:&quot;shave&quot;}]}
+/// {&quot;timestamp&quot;:&quot;2022-02-15T18:47:10.822006Z&quot;,&quot;level&quot;:&quot;DEBUG&quot;,&quot;fields&quot;:{&quot;yak&quot;:3,&quot;shaved&quot;:false},&quot;target&quot;:&quot;yak_events&quot;,&quot;spans&quot;:[{&quot;yaks&quot;:3,&quot;name&quot;:&quot;shaving_yaks&quot;}]}
+/// {&quot;timestamp&quot;:&quot;2022-02-15T18:47:10.822041Z&quot;,&quot;level&quot;:&quot;ERROR&quot;,&quot;fields&quot;:{&quot;message&quot;:&quot;failed to shave yak&quot;,&quot;yak&quot;:3,&quot;error&quot;:&quot;missing yak&quot;},&quot;target&quot;:&quot;fmt_json::yak_shave&quot;,&quot;spans&quot;:[{&quot;yaks&quot;:3,&quot;name&quot;:&quot;shaving_yaks&quot;}]}
+/// {&quot;timestamp&quot;:&quot;2022-02-15T18:47:10.822079Z&quot;,&quot;level&quot;:&quot;TRACE&quot;,&quot;fields&quot;:{&quot;yaks_shaved&quot;:2},&quot;target&quot;:&quot;fmt_json::yak_shave&quot;,&quot;spans&quot;:[{&quot;yaks&quot;:3,&quot;name&quot;:&quot;shaving_yaks&quot;}]}
+/// {&quot;timestamp&quot;:&quot;2022-02-15T18:47:10.822117Z&quot;,&quot;level&quot;:&quot;INFO&quot;,&quot;fields&quot;:{&quot;message&quot;:&quot;yak shaving completed&quot;,&quot;all_yaks_shaved&quot;:false},&quot;target&quot;:&quot;fmt_json&quot;}
+/// </pre>
+///
+/// # Options
+///
+/// This formatter exposes additional options to configure the structure of the
+/// output JSON objects:
+///
+/// - [`Json::flatten_event`] can be used to enable flattening event fields into
+///   the root
+/// - [`Json::with_current_span`] can be used to control logging of the current
+///   span
+/// - [`Json::with_span_list`] can be used to control logging of the span list
+///   object.
+///
+/// By default, event fields are not flattened, and both current span and span
+/// list are logged.
+///
+/// # Valuable Support
+///
+/// Experimental support is available for using the [`valuable`] crate to record
+/// user-defined values as structured JSON. When the ["valuable" unstable
+/// feature][unstable] is enabled, types implementing [`valuable::Valuable`] will
+/// be recorded as structured JSON, rather than
+/// using their [`std::fmt::Debug`] implementations.
+///
+/// **Note**: This is an experimental feature. [Unstable features][unstable]
+/// must be enabled in order to use `valuable` support.
+///
+/// [`Json::flatten_event`]: Json::flatten_event()
+/// [`Json::with_current_span`]: Json::with_current_span()
+/// [`Json::with_span_list`]: Json::with_span_list()
+/// [`valuable`]: https://crates.io/crates/valuable
+/// [unstable]: crate#unstable-features
+/// [`valuable::Valuable`]: https://docs.rs/valuable/latest/valuable/trait.Valuable.html
+#[derive(Debug, Copy, Clone, Eq, PartialEq)]
+pub struct Json {
+    pub(crate) flatten_event: bool,
+    pub(crate) display_current_span: bool,
+    pub(crate) display_span_list: bool,
+}
+
+impl Json {
+    /// If set to `true` event metadata will be flattened into the root object.
+    pub fn flatten_event(&mut self, flatten_event: bool) {
+        self.flatten_event = flatten_event;
+    }
+
+    /// If set to `false`, formatted events won't contain a field for the current span.
+    pub fn with_current_span(&mut self, display_current_span: bool) {
+        self.display_current_span = display_current_span;
+    }
+
+    /// If set to `false`, formatted events won't contain a list of all currently
+    /// entered spans. Spans are logged in a list from root to leaf.
+    pub fn with_span_list(&mut self, display_span_list: bool) {
+        self.display_span_list = display_span_list;
+    }
+}
+
+struct SerializableContext<'a, 'b, Span, N>(
+    &'b crate::layer::Context<'a, Span>,
+    std::marker::PhantomData<N>,
+)
+where
+    Span: Subscriber + for<'lookup> crate::registry::LookupSpan<'lookup>,
+    N: for<'writer> FormatFields<'writer> + 'static;
+
+impl<Span, N> serde::ser::Serialize for SerializableContext<'_, '_, Span, N>
+where
+    Span: Subscriber + for<'lookup> crate::registry::LookupSpan<'lookup>,
+    N: for<'writer> FormatFields<'writer> + 'static,
+{
+    fn serialize<Ser>(&self, serializer_o: Ser) -> Result<Ser::Ok, Ser::Error>
+    where
+        Ser: serde::ser::Serializer,
+    {
+        use serde::ser::SerializeSeq;
+        let mut serializer = serializer_o.serialize_seq(None)?;
+
+        if let Some(leaf_span) = self.0.lookup_current() {
+            for span in leaf_span.scope().from_root() {
+                serializer.serialize_element(&SerializableSpan(&span, self.1))?;
+            }
+        }
+
+        serializer.end()
+    }
+}
+
+struct SerializableSpan<'a, 'b, Span, N>(
+    &'b crate::registry::SpanRef<'a, Span>,
+    std::marker::PhantomData<N>,
+)
+where
+    Span: for<'lookup> crate::registry::LookupSpan<'lookup>,
+    N: for<'writer> FormatFields<'writer> + 'static;
+
+impl<Span, N> serde::ser::Serialize for SerializableSpan<'_, '_, Span, N>
+where
+    Span: for<'lookup> crate::registry::LookupSpan<'lookup>,
+    N: for<'writer> FormatFields<'writer> + 'static,
+{
+    fn serialize<Ser>(&self, serializer: Ser) -> Result<Ser::Ok, Ser::Error>
+    where
+        Ser: serde::ser::Serializer,
+    {
+        let mut serializer = serializer.serialize_map(None)?;
+
+        let ext = self.0.extensions();
+        let data = ext
+            .get::<FormattedFields<N>>()
+            .expect("Unable to find FormattedFields in extensions; this is a bug");
+
+        // TODO: let's _not_ do this, but this resolves
+        // https://github.com/tokio-rs/tracing/issues/391.
+        // We should probably rework this to use a `serde_json::Value` or something
+        // similar in a JSON-specific layer, but I'd (david)
+        // rather have a uglier fix now rather than shipping broken JSON.
+        match serde_json::from_str::<serde_json::Value>(data) {
+            Ok(serde_json::Value::Object(fields)) => {
+                for field in fields {
+                    serializer.serialize_entry(&field.0, &field.1)?;
+                }
+            }
+            // We have fields for this span which are valid JSON but not an object.
+            // This is probably a bug, so panic if we're in debug mode
+            Ok(_) if cfg!(debug_assertions) => panic!(
+                "span '{}' had malformed fields! this is a bug.\n  error: invalid JSON object\n  fields: {:?}",
+                self.0.metadata().name(),
+                data
+            ),
+            // If we *aren't* in debug mode, it's probably best not to
+            // crash the program, let's log the field found but also an
+            // message saying it's type  is invalid
+            Ok(value) => {
+                serializer.serialize_entry("field", &value)?;
+                serializer.serialize_entry("field_error", "field was no a valid object")?
+            }
+            // We have previously recorded fields for this span
+            // should be valid JSON. However, they appear to *not*
+            // be valid JSON. This is almost certainly a bug, so
+            // panic if we're in debug mode
+            Err(e) if cfg!(debug_assertions) => panic!(
+                "span '{}' had malformed fields! this is a bug.\n  error: {}\n  fields: {:?}",
+                self.0.metadata().name(),
+                e,
+                data
+            ),
+            // If we *aren't* in debug mode, it's probably best not
+            // crash the program, but let's at least make sure it's clear
+            // that the fields are not supposed to be missing.
+            Err(e) => serializer.serialize_entry("field_error", &format!("{}", e))?,
+        };
+        serializer.serialize_entry("name", self.0.metadata().name())?;
+        serializer.end()
+    }
+}
+
+impl<S, N, T> FormatEvent<S, N> for Format<Json, T>
+where
+    S: Subscriber + for<'lookup> LookupSpan<'lookup>,
+    N: for<'writer> FormatFields<'writer> + 'static,
+    T: FormatTime,
+{
+    fn format_event(
+        &self,
+        ctx: &FmtContext<'_, S, N>,
+        mut writer: Writer<'_>,
+        event: &Event<'_>,
+    ) -> fmt::Result
+    where
+        S: Subscriber + for<'a> LookupSpan<'a>,
+    {
+        let mut timestamp = String::new();
+        self.timer.format_time(&mut Writer::new(&mut timestamp))?;
+
+        #[cfg(feature = "tracing-log")]
+        let normalized_meta = event.normalized_metadata();
+        #[cfg(feature = "tracing-log")]
+        let meta = normalized_meta.as_ref().unwrap_or_else(|| event.metadata());
+        #[cfg(not(feature = "tracing-log"))]
+        let meta = event.metadata();
+
+        let mut visit = || {
+            let mut serializer = Serializer::new(WriteAdaptor::new(&mut writer));
+
+            let mut serializer = serializer.serialize_map(None)?;
+
+            if self.display_timestamp {
+                serializer.serialize_entry("timestamp", &timestamp)?;
+            }
+
+            if self.display_level {
+                serializer.serialize_entry("level", &meta.level().as_serde())?;
+            }
+
+            let format_field_marker: std::marker::PhantomData<N> = std::marker::PhantomData;
+
+            let current_span = if self.format.display_current_span || self.format.display_span_list
+            {
+                event
+                    .parent()
+                    .and_then(|id| ctx.span(id))
+                    .or_else(|| ctx.lookup_current())
+            } else {
+                None
+            };
+
+            if self.format.flatten_event {
+                let mut visitor = tracing_serde::SerdeMapVisitor::new(serializer);
+                event.record(&mut visitor);
+
+                serializer = visitor.take_serializer()?;
+            } else {
+                use tracing_serde::fields::AsMap;
+                serializer.serialize_entry("fields", &event.field_map())?;
+            };
+
+            if self.display_target {
+                serializer.serialize_entry("target", meta.target())?;
+            }
+
+            if self.display_filename {
+                if let Some(filename) = meta.file() {
+                    serializer.serialize_entry("filename", filename)?;
+                }
+            }
+
+            if self.display_line_number {
+                if let Some(line_number) = meta.line() {
+                    serializer.serialize_entry("line_number", &line_number)?;
+                }
+            }
+
+            if self.format.display_current_span {
+                if let Some(ref span) = current_span {
+                    serializer
+                        .serialize_entry("span", &SerializableSpan(span, format_field_marker))
+                        .unwrap_or(());
+                }
+            }
+
+            if self.format.display_span_list && current_span.is_some() {
+                serializer.serialize_entry(
+                    "spans",
+                    &SerializableContext(&ctx.ctx, format_field_marker),
+                )?;
+            }
+
+            if self.display_thread_name {
+                let current_thread = std::thread::current();
+                match current_thread.name() {
+                    Some(name) => {
+                        serializer.serialize_entry("threadName", name)?;
+                    }
+                    // fall-back to thread id when name is absent and ids are not enabled
+                    None if !self.display_thread_id => {
+                        serializer
+                            .serialize_entry("threadName", &format!("{:?}", current_thread.id()))?;
+                    }
+                    _ => {}
+                }
+            }
+
+            if self.display_thread_id {
+                serializer
+                    .serialize_entry("threadId", &format!("{:?}", std::thread::current().id()))?;
+            }
+
+            serializer.end()
+        };
+
+        visit().map_err(|_| fmt::Error)?;
+        writeln!(writer)
+    }
+}
+
+impl Default for Json {
+    fn default() -> Json {
+        Json {
+            flatten_event: false,
+            display_current_span: true,
+            display_span_list: true,
+        }
+    }
+}
+
+/// The JSON [`FormatFields`] implementation.
+///
+#[derive(Debug)]
+pub struct JsonFields {
+    // reserve the ability to add fields to this without causing a breaking
+    // change in the future.
+    _private: (),
+}
+
+impl JsonFields {
+    /// Returns a new JSON [`FormatFields`] implementation.
+    ///
+    pub fn new() -> Self {
+        Self { _private: () }
+    }
+}
+
+impl Default for JsonFields {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
+impl<'a> FormatFields<'a> for JsonFields {
+    /// Format the provided `fields` to the provided `writer`, returning a result.
+    fn format_fields<R: RecordFields>(&self, mut writer: Writer<'_>, fields: R) -> fmt::Result {
+        let mut v = JsonVisitor::new(&mut writer);
+        fields.record(&mut v);
+        v.finish()
+    }
+
+    /// Record additional field(s) on an existing span.
+    ///
+    /// By default, this appends a space to the current set of fields if it is
+    /// non-empty, and then calls `self.format_fields`. If different behavior is
+    /// required, the default implementation of this method can be overridden.
+    fn add_fields(
+        &self,
+        current: &'a mut FormattedFields<Self>,
+        fields: &Record<'_>,
+    ) -> fmt::Result {
+        if current.is_empty() {
+            // If there are no previously recorded fields, we can just reuse the
+            // existing string.
+            let mut writer = current.as_writer();
+            let mut v = JsonVisitor::new(&mut writer);
+            fields.record(&mut v);
+            v.finish()?;
+            return Ok(());
+        }
+
+        // If fields were previously recorded on this span, we need to parse
+        // the current set of fields as JSON, add the new fields, and
+        // re-serialize them. Otherwise, if we just appended the new fields
+        // to a previously serialized JSON object, we would end up with
+        // malformed JSON.
+        //
+        // XXX(eliza): this is far from efficient, but unfortunately, it is
+        // necessary as long as the JSON formatter is implemented on top of
+        // an interface that stores all formatted fields as strings.
+        //
+        // We should consider reimplementing the JSON formatter as a
+        // separate layer, rather than a formatter for the `fmt` layer —
+        // then, we could store fields as JSON values, and add to them
+        // without having to parse and re-serialize.
+        let mut new = String::new();
+        let map: BTreeMap<&'_ str, serde_json::Value> =
+            serde_json::from_str(current).map_err(|_| fmt::Error)?;
+        let mut v = JsonVisitor::new(&mut new);
+        v.values = map;
+        fields.record(&mut v);
+        v.finish()?;
+        current.fields = new;
+
+        Ok(())
+    }
+}
+
+/// The [visitor] produced by [`JsonFields`]'s [`MakeVisitor`] implementation.
+///
+/// [visitor]: crate::field::Visit
+/// [`MakeVisitor`]: crate::field::MakeVisitor
+pub struct JsonVisitor<'a> {
+    values: BTreeMap<&'a str, serde_json::Value>,
+    writer: &'a mut dyn Write,
+}
+
+impl fmt::Debug for JsonVisitor<'_> {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        f.write_fmt(format_args!("JsonVisitor {{ values: {:?} }}", self.values))
+    }
+}
+
+impl<'a> JsonVisitor<'a> {
+    /// Returns a new default visitor that formats to the provided `writer`.
+    ///
+    /// # Arguments
+    /// - `writer`: the writer to format to.
+    /// - `is_empty`: whether or not any fields have been previously written to
+    ///   that writer.
+    pub fn new(writer: &'a mut dyn Write) -> Self {
+        Self {
+            values: BTreeMap::new(),
+            writer,
+        }
+    }
+}
+
+impl crate::field::VisitFmt for JsonVisitor<'_> {
+    fn writer(&mut self) -> &mut dyn fmt::Write {
+        self.writer
+    }
+}
+
+impl crate::field::VisitOutput<fmt::Result> for JsonVisitor<'_> {
+    fn finish(self) -> fmt::Result {
+        let inner = || {
+            let mut serializer = Serializer::new(WriteAdaptor::new(self.writer));
+            let mut ser_map = serializer.serialize_map(None)?;
+
+            for (k, v) in self.values {
+                ser_map.serialize_entry(k, &v)?;
+            }
+
+            ser_map.end()
+        };
+
+        if inner().is_err() {
+            Err(fmt::Error)
+        } else {
+            Ok(())
+        }
+    }
+}
+
+impl field::Visit for JsonVisitor<'_> {
+    #[cfg(all(tracing_unstable, feature = "valuable"))]
+    fn record_value(&mut self, field: &Field, value: valuable_crate::Value<'_>) {
+        let value = match serde_json::to_value(valuable_serde::Serializable::new(value)) {
+            Ok(value) => value,
+            Err(_e) => {
+                #[cfg(debug_assertions)]
+                unreachable!(
+                    "`valuable::Valuable` implementations should always serialize \
+                    successfully, but an error occurred: {}",
+                    _e,
+                );
+
+                #[cfg(not(debug_assertions))]
+                return;
+            }
+        };
+
+        self.values.insert(field.name(), value);
+    }
+
+    /// Visit a double precision floating point value.
+    fn record_f64(&mut self, field: &Field, value: f64) {
+        self.values
+            .insert(field.name(), serde_json::Value::from(value));
+    }
+
+    /// Visit a signed 64-bit integer value.
+    fn record_i64(&mut self, field: &Field, value: i64) {
+        self.values
+            .insert(field.name(), serde_json::Value::from(value));
+    }
+
+    /// Visit an unsigned 64-bit integer value.
+    fn record_u64(&mut self, field: &Field, value: u64) {
+        self.values
+            .insert(field.name(), serde_json::Value::from(value));
+    }
+
+    /// Visit a boolean value.
+    fn record_bool(&mut self, field: &Field, value: bool) {
+        self.values
+            .insert(field.name(), serde_json::Value::from(value));
+    }
+
+    /// Visit a string value.
+    fn record_str(&mut self, field: &Field, value: &str) {
+        self.values
+            .insert(field.name(), serde_json::Value::from(value));
+    }
+
+    fn record_bytes(&mut self, field: &Field, value: &[u8]) {
+        self.values
+            .insert(field.name(), serde_json::Value::from(value));
+    }
+
+    fn record_debug(&mut self, field: &Field, value: &dyn fmt::Debug) {
+        match field.name() {
+            // Skip fields that are actually log metadata that have already been handled
+            #[cfg(feature = "tracing-log")]
+            name if name.starts_with("log.") => (),
+            name if name.starts_with("r#") => {
+                self.values
+                    .insert(&name[2..], serde_json::Value::from(format!("{:?}", value)));
+            }
+            name => {
+                self.values
+                    .insert(name, serde_json::Value::from(format!("{:?}", value)));
+            }
+        };
+    }
+}
+#[cfg(test)]
+mod test {
+    use super::*;
+    use crate::fmt::{format::FmtSpan, test::MockMakeWriter, time::FormatTime, SubscriberBuilder};
+    use tracing::{self, subscriber::with_default};
+
+    use std::fmt;
+    use std::path::Path;
+
+    struct MockTime;
+    impl FormatTime for MockTime {
+        fn format_time(&self, w: &mut Writer<'_>) -> fmt::Result {
+            write!(w, "fake time")
+        }
+    }
+
+    fn subscriber() -> SubscriberBuilder<JsonFields, Format<Json>> {
+        SubscriberBuilder::default().json()
+    }
+
+    #[test]
+    fn json() {
+        let expected =
+        "{\"timestamp\":\"fake time\",\"level\":\"INFO\",\"span\":{\"answer\":42,\"name\":\"json_span\",\"number\":3,\"slice\":[97,98,99]},\"spans\":[{\"answer\":42,\"name\":\"json_span\",\"number\":3,\"slice\":[97,98,99]}],\"target\":\"tracing_subscriber::fmt::format::json::test\",\"fields\":{\"message\":\"some json test\"}}\n";
+        let subscriber = subscriber()
+            .flatten_event(false)
+            .with_current_span(true)
+            .with_span_list(true);
+        test_json(expected, subscriber, || {
+            let span = tracing::span!(
+                tracing::Level::INFO,
+                "json_span",
+                answer = 42,
+                number = 3,
+                slice = &b"abc"[..]
+            );
+            let _guard = span.enter();
+            tracing::info!("some json test");
+        });
+    }
+
+    #[test]
+    fn json_filename() {
+        let current_path = Path::new("tracing-subscriber")
+            .join("src")
+            .join("fmt")
+            .join("format")
+            .join("json.rs")
+            .to_str()
+            .expect("path must be valid unicode")
+            // escape windows backslashes
+            .replace('\\', "\\\\");
+        let expected =
+            &format!("{}{}{}",
+                    "{\"timestamp\":\"fake time\",\"level\":\"INFO\",\"span\":{\"answer\":42,\"name\":\"json_span\",\"number\":3},\"spans\":[{\"answer\":42,\"name\":\"json_span\",\"number\":3}],\"target\":\"tracing_subscriber::fmt::format::json::test\",\"filename\":\"",
+                    current_path,
+                    "\",\"fields\":{\"message\":\"some json test\"}}\n");
+        let subscriber = subscriber()
+            .flatten_event(false)
+            .with_current_span(true)
+            .with_file(true)
+            .with_span_list(true);
+        test_json(expected, subscriber, || {
+            let span = tracing::span!(tracing::Level::INFO, "json_span", answer = 42, number = 3);
+            let _guard = span.enter();
+            tracing::info!("some json test");
+        });
+    }
+
+    #[test]
+    fn json_line_number() {
+        let expected =
+            "{\"timestamp\":\"fake time\",\"level\":\"INFO\",\"span\":{\"answer\":42,\"name\":\"json_span\",\"number\":3},\"spans\":[{\"answer\":42,\"name\":\"json_span\",\"number\":3}],\"target\":\"tracing_subscriber::fmt::format::json::test\",\"line_number\":42,\"fields\":{\"message\":\"some json test\"}}\n";
+        let subscriber = subscriber()
+            .flatten_event(false)
+            .with_current_span(true)
+            .with_line_number(true)
+            .with_span_list(true);
+        test_json_with_line_number(expected, subscriber, || {
+            let span = tracing::span!(tracing::Level::INFO, "json_span", answer = 42, number = 3);
+            let _guard = span.enter();
+            tracing::info!("some json test");
+        });
+    }
+
+    #[test]
+    fn json_flattened_event() {
+        let expected =
+        "{\"timestamp\":\"fake time\",\"level\":\"INFO\",\"span\":{\"answer\":42,\"name\":\"json_span\",\"number\":3},\"spans\":[{\"answer\":42,\"name\":\"json_span\",\"number\":3}],\"target\":\"tracing_subscriber::fmt::format::json::test\",\"message\":\"some json test\"}\n";
+
+        let subscriber = subscriber()
+            .flatten_event(true)
+            .with_current_span(true)
+            .with_span_list(true);
+        test_json(expected, subscriber, || {
+            let span = tracing::span!(tracing::Level::INFO, "json_span", answer = 42, number = 3);
+            let _guard = span.enter();
+            tracing::info!("some json test");
+        });
+    }
+
+    #[test]
+    fn json_disabled_current_span_event() {
+        let expected =
+        "{\"timestamp\":\"fake time\",\"level\":\"INFO\",\"spans\":[{\"answer\":42,\"name\":\"json_span\",\"number\":3}],\"target\":\"tracing_subscriber::fmt::format::json::test\",\"fields\":{\"message\":\"some json test\"}}\n";
+        let subscriber = subscriber()
+            .flatten_event(false)
+            .with_current_span(false)
+            .with_span_list(true);
+        test_json(expected, subscriber, || {
+            let span = tracing::span!(tracing::Level::INFO, "json_span", answer = 42, number = 3);
+            let _guard = span.enter();
+            tracing::info!("some json test");
+        });
+    }
+
+    #[test]
+    fn json_disabled_span_list_event() {
+        let expected =
+        "{\"timestamp\":\"fake time\",\"level\":\"INFO\",\"span\":{\"answer\":42,\"name\":\"json_span\",\"number\":3},\"target\":\"tracing_subscriber::fmt::format::json::test\",\"fields\":{\"message\":\"some json test\"}}\n";
+        let subscriber = subscriber()
+            .flatten_event(false)
+            .with_current_span(true)
+            .with_span_list(false);
+        test_json(expected, subscriber, || {
+            let span = tracing::span!(tracing::Level::INFO, "json_span", answer = 42, number = 3);
+            let _guard = span.enter();
+            tracing::info!("some json test");
+        });
+    }
+
+    #[test]
+    fn json_nested_span() {
+        let expected =
+        "{\"timestamp\":\"fake time\",\"level\":\"INFO\",\"span\":{\"answer\":43,\"name\":\"nested_json_span\",\"number\":4},\"spans\":[{\"answer\":42,\"name\":\"json_span\",\"number\":3},{\"answer\":43,\"name\":\"nested_json_span\",\"number\":4}],\"target\":\"tracing_subscriber::fmt::format::json::test\",\"fields\":{\"message\":\"some json test\"}}\n";
+        let subscriber = subscriber()
+            .flatten_event(false)
+            .with_current_span(true)
+            .with_span_list(true);
+        test_json(expected, subscriber, || {
+            let span = tracing::span!(tracing::Level::INFO, "json_span", answer = 42, number = 3);
+            let _guard = span.enter();
+            let span = tracing::span!(
+                tracing::Level::INFO,
+                "nested_json_span",
+                answer = 43,
+                number = 4
+            );
+            let _guard = span.enter();
+            tracing::info!("some json test");
+        });
+    }
+
+    #[test]
+    fn json_no_span() {
+        let expected =
+        "{\"timestamp\":\"fake time\",\"level\":\"INFO\",\"target\":\"tracing_subscriber::fmt::format::json::test\",\"fields\":{\"message\":\"some json test\"}}\n";
+        let subscriber = subscriber()
+            .flatten_event(false)
+            .with_current_span(true)
+            .with_span_list(true);
+        test_json(expected, subscriber, || {
+            tracing::info!("some json test");
+        });
+    }
+
+    #[test]
+    fn record_works() {
+        // This test reproduces issue #707, where using `Span::record` causes
+        // any events inside the span to be ignored.
+
+        let make_writer = MockMakeWriter::default();
+        let subscriber = crate::fmt()
+            .json()
+            .with_writer(make_writer.clone())
+            .finish();
+
+        with_default(subscriber, || {
+            tracing::info!("an event outside the root span");
+            assert_eq!(
+                parse_as_json(&make_writer)["fields"]["message"],
+                "an event outside the root span"
+            );
+
+            let span = tracing::info_span!("the span", na = tracing::field::Empty);
+            span.record("na", "value");
+            let _enter = span.enter();
+
+            tracing::info!("an event inside the root span");
+            assert_eq!(
+                parse_as_json(&make_writer)["fields"]["message"],
+                "an event inside the root span"
+            );
+        });
+    }
+
+    #[test]
+    fn json_span_event_show_correct_context() {
+        let buffer = MockMakeWriter::default();
+        let subscriber = subscriber()
+            .with_writer(buffer.clone())
+            .flatten_event(false)
+            .with_current_span(true)
+            .with_span_list(false)
+            .with_span_events(FmtSpan::FULL)
+            .finish();
+
+        with_default(subscriber, || {
+            let context = "parent";
+            let parent_span = tracing::info_span!("parent_span", context);
+
+            let event = parse_as_json(&buffer);
+            assert_eq!(event["fields"]["message"], "new");
+            assert_eq!(event["span"]["context"], "parent");
+
+            let _parent_enter = parent_span.enter();
+            let event = parse_as_json(&buffer);
+            assert_eq!(event["fields"]["message"], "enter");
+            assert_eq!(event["span"]["context"], "parent");
+
+            let context = "child";
+            let child_span = tracing::info_span!("child_span", context);
+            let event = parse_as_json(&buffer);
+            assert_eq!(event["fields"]["message"], "new");
+            assert_eq!(event["span"]["context"], "child");
+
+            let _child_enter = child_span.enter();
+            let event = parse_as_json(&buffer);
+            assert_eq!(event["fields"]["message"], "enter");
+            assert_eq!(event["span"]["context"], "child");
+
+            drop(_child_enter);
+            let event = parse_as_json(&buffer);
+            assert_eq!(event["fields"]["message"], "exit");
+            assert_eq!(event["span"]["context"], "child");
+
+            drop(child_span);
+            let event = parse_as_json(&buffer);
+            assert_eq!(event["fields"]["message"], "close");
+            assert_eq!(event["span"]["context"], "child");
+
+            drop(_parent_enter);
+            let event = parse_as_json(&buffer);
+            assert_eq!(event["fields"]["message"], "exit");
+            assert_eq!(event["span"]["context"], "parent");
+
+            drop(parent_span);
+            let event = parse_as_json(&buffer);
+            assert_eq!(event["fields"]["message"], "close");
+            assert_eq!(event["span"]["context"], "parent");
+        });
+    }
+
+    #[test]
+    fn json_span_event_with_no_fields() {
+        // Check span events serialize correctly.
+        // Discussion: https://github.com/tokio-rs/tracing/issues/829#issuecomment-661984255
+        let buffer = MockMakeWriter::default();
+        let subscriber = subscriber()
+            .with_writer(buffer.clone())
+            .flatten_event(false)
+            .with_current_span(false)
+            .with_span_list(false)
+            .with_span_events(FmtSpan::FULL)
+            .finish();
+
+        with_default(subscriber, || {
+            let span = tracing::info_span!("valid_json");
+            assert_eq!(parse_as_json(&buffer)["fields"]["message"], "new");
+
+            let _enter = span.enter();
+            assert_eq!(parse_as_json(&buffer)["fields"]["message"], "enter");
+
+            drop(_enter);
+            assert_eq!(parse_as_json(&buffer)["fields"]["message"], "exit");
+
+            drop(span);
+            assert_eq!(parse_as_json(&buffer)["fields"]["message"], "close");
+        });
+    }
+
+    fn parse_as_json(buffer: &MockMakeWriter) -> serde_json::Value {
+        let buf = String::from_utf8(buffer.buf().to_vec()).unwrap();
+        let json = buf
+            .lines()
+            .last()
+            .expect("expected at least one line to be written!");
+        match serde_json::from_str(json) {
+            Ok(v) => v,
+            Err(e) => panic!(
+                "assertion failed: JSON shouldn't be malformed\n  error: {}\n  json: {}",
+                e, json
+            ),
+        }
+    }
+
+    fn test_json<T>(
+        expected: &str,
+        builder: crate::fmt::SubscriberBuilder<JsonFields, Format<Json>>,
+        producer: impl FnOnce() -> T,
+    ) {
+        let make_writer = MockMakeWriter::default();
+        let subscriber = builder
+            .with_writer(make_writer.clone())
+            .with_timer(MockTime)
+            .finish();
+
+        with_default(subscriber, producer);
+
+        let buf = make_writer.buf();
+        let actual = std::str::from_utf8(&buf[..]).unwrap();
+        assert_eq!(
+            serde_json::from_str::<std::collections::HashMap<&str, serde_json::Value>>(expected)
+                .unwrap(),
+            serde_json::from_str(actual).unwrap()
+        );
+    }
+
+    fn test_json_with_line_number<T>(
+        expected: &str,
+        builder: crate::fmt::SubscriberBuilder<JsonFields, Format<Json>>,
+        producer: impl FnOnce() -> T,
+    ) {
+        let make_writer = MockMakeWriter::default();
+        let subscriber = builder
+            .with_writer(make_writer.clone())
+            .with_timer(MockTime)
+            .finish();
+
+        with_default(subscriber, producer);
+
+        let buf = make_writer.buf();
+        let actual = std::str::from_utf8(&buf[..]).unwrap();
+        let mut expected =
+            serde_json::from_str::<std::collections::HashMap<&str, serde_json::Value>>(expected)
+                .unwrap();
+        let expect_line_number = expected.remove("line_number").is_some();
+        let mut actual: std::collections::HashMap<&str, serde_json::Value> =
+            serde_json::from_str(actual).unwrap();
+        let line_number = actual.remove("line_number");
+        if expect_line_number {
+            assert_eq!(line_number.map(|x| x.is_number()), Some(true));
+        } else {
+            assert!(line_number.is_none());
+        }
+        assert_eq!(actual, expected);
+    }
+}

vendor/tracing-subscriber/src/fmt/format/pretty.rs

@@ -0,0 +1,514 @@
+use super::*;
+use crate::{
+    field::{VisitFmt, VisitOutput},
+    fmt::fmt_layer::{FmtContext, FormattedFields},
+    registry::LookupSpan,
+};
+
+use std::fmt;
+use tracing_core::{
+    field::{self, Field},
+    Event, Level, Subscriber,
+};
+
+#[cfg(feature = "tracing-log")]
+use tracing_log::NormalizeEvent;
+
+use nu_ansi_term::{Color, Style};
+
+/// An excessively pretty, human-readable event formatter.
+///
+/// Unlike the [`Full`], [`Compact`], and [`Json`] formatters, this is a
+/// multi-line output format. Each individual event may output multiple lines of
+/// text.
+///
+/// # Example Output
+///
+/// <pre><font color="#4E9A06"><b>:;</b></font> <font color="#4E9A06">cargo</font> run --example fmt-pretty
+/// <font color="#4E9A06"><b>    Finished</b></font> dev [unoptimized + debuginfo] target(s) in 0.08s
+/// <font color="#4E9A06"><b>     Running</b></font> `target/debug/examples/fmt-pretty`
+///   2022-02-15T18:44:24.535324Z <font color="#4E9A06"> INFO</font> <font color="#4E9A06"><b>fmt_pretty</b></font><font color="#4E9A06">: preparing to shave yaks, </font><font color="#4E9A06"><b>number_of_yaks</b></font><font color="#4E9A06">: 3</font>
+///     <font color="#AAAAAA"><i>at</i></font> examples/examples/fmt-pretty.rs:16 <font color="#AAAAAA"><i>on</i></font> main
+///
+///   2022-02-15T18:44:24.535403Z <font color="#4E9A06"> INFO</font> <font color="#4E9A06"><b>fmt_pretty::yak_shave</b></font><font color="#4E9A06">: shaving yaks</font>
+///     <font color="#AAAAAA"><i>at</i></font> examples/examples/fmt/yak_shave.rs:41 <font color="#AAAAAA"><i>on</i></font> main
+///     <font color="#AAAAAA"><i>in</i></font> fmt_pretty::yak_shave::<b>shaving_yaks</b> <font color="#AAAAAA"><i>with</i></font> <b>yaks</b>: 3
+///
+///   2022-02-15T18:44:24.535442Z <font color="#75507B">TRACE</font> <font color="#75507B"><b>fmt_pretty::yak_shave</b></font><font color="#75507B">: hello! I&apos;m gonna shave a yak, </font><font color="#75507B"><b>excitement</b></font><font color="#75507B">: &quot;yay!&quot;</font>
+///     <font color="#AAAAAA"><i>at</i></font> examples/examples/fmt/yak_shave.rs:16 <font color="#AAAAAA"><i>on</i></font> main
+///     <font color="#AAAAAA"><i>in</i></font> fmt_pretty::yak_shave::<b>shave</b> <font color="#AAAAAA"><i>with</i></font> <b>yak</b>: 1
+///     <font color="#AAAAAA"><i>in</i></font> fmt_pretty::yak_shave::<b>shaving_yaks</b> <font color="#AAAAAA"><i>with</i></font> <b>yaks</b>: 3
+///
+///   2022-02-15T18:44:24.535469Z <font color="#75507B">TRACE</font> <font color="#75507B"><b>fmt_pretty::yak_shave</b></font><font color="#75507B">: yak shaved successfully</font>
+///     <font color="#AAAAAA"><i>at</i></font> examples/examples/fmt/yak_shave.rs:25 <font color="#AAAAAA"><i>on</i></font> main
+///     <font color="#AAAAAA"><i>in</i></font> fmt_pretty::yak_shave::<b>shave</b> <font color="#AAAAAA"><i>with</i></font> <b>yak</b>: 1
+///     <font color="#AAAAAA"><i>in</i></font> fmt_pretty::yak_shave::<b>shaving_yaks</b> <font color="#AAAAAA"><i>with</i></font> <b>yaks</b>: 3
+///
+///   2022-02-15T18:44:24.535502Z <font color="#3465A4">DEBUG</font> <font color="#3465A4"><b>yak_events</b></font><font color="#3465A4">: </font><font color="#3465A4"><b>yak</b></font><font color="#3465A4">: 1, </font><font color="#3465A4"><b>shaved</b></font><font color="#3465A4">: true</font>
+///     <font color="#AAAAAA"><i>at</i></font> examples/examples/fmt/yak_shave.rs:46 <font color="#AAAAAA"><i>on</i></font> main
+///     <font color="#AAAAAA"><i>in</i></font> fmt_pretty::yak_shave::<b>shaving_yaks</b> <font color="#AAAAAA"><i>with</i></font> <b>yaks</b>: 3
+///
+///   2022-02-15T18:44:24.535524Z <font color="#75507B">TRACE</font> <font color="#75507B"><b>fmt_pretty::yak_shave</b></font><font color="#75507B">: </font><font color="#75507B"><b>yaks_shaved</b></font><font color="#75507B">: 1</font>
+///     <font color="#AAAAAA"><i>at</i></font> examples/examples/fmt/yak_shave.rs:55 <font color="#AAAAAA"><i>on</i></font> main
+///     <font color="#AAAAAA"><i>in</i></font> fmt_pretty::yak_shave::<b>shaving_yaks</b> <font color="#AAAAAA"><i>with</i></font> <b>yaks</b>: 3
+///
+///   2022-02-15T18:44:24.535551Z <font color="#75507B">TRACE</font> <font color="#75507B"><b>fmt_pretty::yak_shave</b></font><font color="#75507B">: hello! I&apos;m gonna shave a yak, </font><font color="#75507B"><b>excitement</b></font><font color="#75507B">: &quot;yay!&quot;</font>
+///     <font color="#AAAAAA"><i>at</i></font> examples/examples/fmt/yak_shave.rs:16 <font color="#AAAAAA"><i>on</i></font> main
+///     <font color="#AAAAAA"><i>in</i></font> fmt_pretty::yak_shave::<b>shave</b> <font color="#AAAAAA"><i>with</i></font> <b>yak</b>: 2
+///     <font color="#AAAAAA"><i>in</i></font> fmt_pretty::yak_shave::<b>shaving_yaks</b> <font color="#AAAAAA"><i>with</i></font> <b>yaks</b>: 3
+///
+///   2022-02-15T18:44:24.535573Z <font color="#75507B">TRACE</font> <font color="#75507B"><b>fmt_pretty::yak_shave</b></font><font color="#75507B">: yak shaved successfully</font>
+///     <font color="#AAAAAA"><i>at</i></font> examples/examples/fmt/yak_shave.rs:25 <font color="#AAAAAA"><i>on</i></font> main
+///     <font color="#AAAAAA"><i>in</i></font> fmt_pretty::yak_shave::<b>shave</b> <font color="#AAAAAA"><i>with</i></font> <b>yak</b>: 2
+///     <font color="#AAAAAA"><i>in</i></font> fmt_pretty::yak_shave::<b>shaving_yaks</b> <font color="#AAAAAA"><i>with</i></font> <b>yaks</b>: 3
+///
+///   2022-02-15T18:44:24.535600Z <font color="#3465A4">DEBUG</font> <font color="#3465A4"><b>yak_events</b></font><font color="#3465A4">: </font><font color="#3465A4"><b>yak</b></font><font color="#3465A4">: 2, </font><font color="#3465A4"><b>shaved</b></font><font color="#3465A4">: true</font>
+///     <font color="#AAAAAA"><i>at</i></font> examples/examples/fmt/yak_shave.rs:46 <font color="#AAAAAA"><i>on</i></font> main
+///     <font color="#AAAAAA"><i>in</i></font> fmt_pretty::yak_shave::<b>shaving_yaks</b> <font color="#AAAAAA"><i>with</i></font> <b>yaks</b>: 3
+///
+///   2022-02-15T18:44:24.535618Z <font color="#75507B">TRACE</font> <font color="#75507B"><b>fmt_pretty::yak_shave</b></font><font color="#75507B">: </font><font color="#75507B"><b>yaks_shaved</b></font><font color="#75507B">: 2</font>
+///     <font color="#AAAAAA"><i>at</i></font> examples/examples/fmt/yak_shave.rs:55 <font color="#AAAAAA"><i>on</i></font> main
+///     <font color="#AAAAAA"><i>in</i></font> fmt_pretty::yak_shave::<b>shaving_yaks</b> <font color="#AAAAAA"><i>with</i></font> <b>yaks</b>: 3
+///
+///   2022-02-15T18:44:24.535644Z <font color="#75507B">TRACE</font> <font color="#75507B"><b>fmt_pretty::yak_shave</b></font><font color="#75507B">: hello! I&apos;m gonna shave a yak, </font><font color="#75507B"><b>excitement</b></font><font color="#75507B">: &quot;yay!&quot;</font>
+///     <font color="#AAAAAA"><i>at</i></font> examples/examples/fmt/yak_shave.rs:16 <font color="#AAAAAA"><i>on</i></font> main
+///     <font color="#AAAAAA"><i>in</i></font> fmt_pretty::yak_shave::<b>shave</b> <font color="#AAAAAA"><i>with</i></font> <b>yak</b>: 3
+///     <font color="#AAAAAA"><i>in</i></font> fmt_pretty::yak_shave::<b>shaving_yaks</b> <font color="#AAAAAA"><i>with</i></font> <b>yaks</b>: 3
+///
+///   2022-02-15T18:44:24.535670Z <font color="#C4A000"> WARN</font> <font color="#C4A000"><b>fmt_pretty::yak_shave</b></font><font color="#C4A000">: could not locate yak</font>
+///     <font color="#AAAAAA"><i>at</i></font> examples/examples/fmt/yak_shave.rs:18 <font color="#AAAAAA"><i>on</i></font> main
+///     <font color="#AAAAAA"><i>in</i></font> fmt_pretty::yak_shave::<b>shave</b> <font color="#AAAAAA"><i>with</i></font> <b>yak</b>: 3
+///     <font color="#AAAAAA"><i>in</i></font> fmt_pretty::yak_shave::<b>shaving_yaks</b> <font color="#AAAAAA"><i>with</i></font> <b>yaks</b>: 3
+///
+///   2022-02-15T18:44:24.535698Z <font color="#3465A4">DEBUG</font> <font color="#3465A4"><b>yak_events</b></font><font color="#3465A4">: </font><font color="#3465A4"><b>yak</b></font><font color="#3465A4">: 3, </font><font color="#3465A4"><b>shaved</b></font><font color="#3465A4">: false</font>
+///     <font color="#AAAAAA"><i>at</i></font> examples/examples/fmt/yak_shave.rs:46 <font color="#AAAAAA"><i>on</i></font> main
+///     <font color="#AAAAAA"><i>in</i></font> fmt_pretty::yak_shave::<b>shaving_yaks</b> <font color="#AAAAAA"><i>with</i></font> <b>yaks</b>: 3
+///
+///   2022-02-15T18:44:24.535720Z <font color="#CC0000">ERROR</font> <font color="#CC0000"><b>fmt_pretty::yak_shave</b></font><font color="#CC0000">: failed to shave yak, </font><font color="#CC0000"><b>yak</b></font><font color="#CC0000">: 3, </font><font color="#CC0000"><b>error</b></font><font color="#CC0000">: missing yak, </font><font color="#CC0000"><b>error.sources</b></font><font color="#CC0000">: [out of space, out of cash]</font>
+///     <font color="#AAAAAA"><i>at</i></font> examples/examples/fmt/yak_shave.rs:51 <font color="#AAAAAA"><i>on</i></font> main
+///     <font color="#AAAAAA"><i>in</i></font> fmt_pretty::yak_shave::<b>shaving_yaks</b> <font color="#AAAAAA"><i>with</i></font> <b>yaks</b>: 3
+///
+///   2022-02-15T18:44:24.535742Z <font color="#75507B">TRACE</font> <font color="#75507B"><b>fmt_pretty::yak_shave</b></font><font color="#75507B">: </font><font color="#75507B"><b>yaks_shaved</b></font><font color="#75507B">: 2</font>
+///     <font color="#AAAAAA"><i>at</i></font> examples/examples/fmt/yak_shave.rs:55 <font color="#AAAAAA"><i>on</i></font> main
+///     <font color="#AAAAAA"><i>in</i></font> fmt_pretty::yak_shave::<b>shaving_yaks</b> <font color="#AAAAAA"><i>with</i></font> <b>yaks</b>: 3
+///
+///   2022-02-15T18:44:24.535765Z <font color="#4E9A06"> INFO</font> <font color="#4E9A06"><b>fmt_pretty</b></font><font color="#4E9A06">: yak shaving completed, </font><font color="#4E9A06"><b>all_yaks_shaved</b></font><font color="#4E9A06">: false</font>
+///     <font color="#AAAAAA"><i>at</i></font> examples/examples/fmt-pretty.rs:19 <font color="#AAAAAA"><i>on</i></font> main
+/// </pre>
+#[derive(Debug, Clone, Eq, PartialEq)]
+pub struct Pretty {
+    display_location: bool,
+}
+
+/// The [visitor] produced by [`Pretty`]'s [`MakeVisitor`] implementation.
+///
+/// [visitor]: field::Visit
+/// [`MakeVisitor`]: crate::field::MakeVisitor
+#[derive(Debug)]
+pub struct PrettyVisitor<'a> {
+    writer: Writer<'a>,
+    is_empty: bool,
+    style: Style,
+    result: fmt::Result,
+}
+
+/// An excessively pretty, human-readable [`MakeVisitor`] implementation.
+///
+/// [`MakeVisitor`]: crate::field::MakeVisitor
+#[derive(Debug)]
+pub struct PrettyFields {
+    /// A value to override the provided `Writer`'s ANSI formatting
+    /// configuration.
+    ///
+    /// If this is `Some`, we override the `Writer`'s ANSI setting. This is
+    /// necessary in order to continue supporting the deprecated
+    /// `PrettyFields::with_ansi` method. If it is `None`, we don't override the
+    /// ANSI formatting configuration (because the deprecated method was not
+    /// called).
+    // TODO: when `PrettyFields::with_ansi` is removed, we can get rid
+    // of this entirely.
+    ansi: Option<bool>,
+}
+
+// === impl Pretty ===
+
+impl Default for Pretty {
+    fn default() -> Self {
+        Self {
+            display_location: true,
+        }
+    }
+}
+
+impl Pretty {
+    fn style_for(level: &Level) -> Style {
+        match *level {
+            Level::TRACE => Style::new().fg(Color::Purple),
+            Level::DEBUG => Style::new().fg(Color::Blue),
+            Level::INFO => Style::new().fg(Color::Green),
+            Level::WARN => Style::new().fg(Color::Yellow),
+            Level::ERROR => Style::new().fg(Color::Red),
+        }
+    }
+
+    /// Sets whether the event's source code location is displayed.
+    ///
+    /// This defaults to `true`.
+    #[deprecated(
+        since = "0.3.6",
+        note = "all formatters now support configurable source locations. Use `Format::with_source_location` instead."
+    )]
+    pub fn with_source_location(self, display_location: bool) -> Self {
+        Self {
+            display_location,
+            ..self
+        }
+    }
+}
+
+impl<C, N, T> FormatEvent<C, N> for Format<Pretty, T>
+where
+    C: Subscriber + for<'a> LookupSpan<'a>,
+    N: for<'a> FormatFields<'a> + 'static,
+    T: FormatTime,
+{
+    fn format_event(
+        &self,
+        ctx: &FmtContext<'_, C, N>,
+        mut writer: Writer<'_>,
+        event: &Event<'_>,
+    ) -> fmt::Result {
+        #[cfg(feature = "tracing-log")]
+        let normalized_meta = event.normalized_metadata();
+        #[cfg(feature = "tracing-log")]
+        let meta = normalized_meta.as_ref().unwrap_or_else(|| event.metadata());
+        #[cfg(not(feature = "tracing-log"))]
+        let meta = event.metadata();
+        write!(&mut writer, "  ")?;
+
+        // if the `Format` struct *also* has an ANSI color configuration,
+        // override the writer...the API for configuring ANSI color codes on the
+        // `Format` struct is deprecated, but we still need to honor those
+        // configurations.
+        if let Some(ansi) = self.ansi {
+            writer = writer.with_ansi(ansi);
+        }
+
+        self.format_timestamp(&mut writer)?;
+
+        let style = if self.display_level && writer.has_ansi_escapes() {
+            Pretty::style_for(meta.level())
+        } else {
+            Style::new()
+        };
+
+        if self.display_level {
+            write!(
+                writer,
+                "{} ",
+                super::FmtLevel::new(meta.level(), writer.has_ansi_escapes())
+            )?;
+        }
+
+        if self.display_target {
+            let target_style = if writer.has_ansi_escapes() {
+                style.bold()
+            } else {
+                style
+            };
+            write!(
+                writer,
+                "{}{}{}:",
+                target_style.prefix(),
+                meta.target(),
+                target_style.infix(style)
+            )?;
+        }
+        let line_number = if self.display_line_number {
+            meta.line()
+        } else {
+            None
+        };
+
+        // If the file name is disabled, format the line number right after the
+        // target. Otherwise, if we also display the file, it'll go on a
+        // separate line.
+        if let (Some(line_number), false, true) = (
+            line_number,
+            self.display_filename,
+            self.format.display_location,
+        ) {
+            write!(
+                writer,
+                "{}{}{}:",
+                style.prefix(),
+                line_number,
+                style.infix(style)
+            )?;
+        }
+
+        writer.write_char(' ')?;
+
+        let mut v = PrettyVisitor::new(writer.by_ref(), true).with_style(style);
+        event.record(&mut v);
+        v.finish()?;
+        writer.write_char('\n')?;
+
+        let dimmed = if writer.has_ansi_escapes() {
+            Style::new().dimmed().italic()
+        } else {
+            Style::new()
+        };
+        let thread = self.display_thread_name || self.display_thread_id;
+
+        if let (Some(file), true, true) = (
+            meta.file(),
+            self.format.display_location,
+            self.display_filename,
+        ) {
+            write!(writer, "    {} {}", dimmed.paint("at"), file,)?;
+
+            if let Some(line) = line_number {
+                write!(writer, ":{}", line)?;
+            }
+            writer.write_char(if thread { ' ' } else { '\n' })?;
+        } else if thread {
+            write!(writer, "    ")?;
+        };
+
+        if thread {
+            write!(writer, "{} ", dimmed.paint("on"))?;
+            let thread = std::thread::current();
+            if self.display_thread_name {
+                if let Some(name) = thread.name() {
+                    write!(writer, "{}", name)?;
+                    if self.display_thread_id {
+                        writer.write_char(' ')?;
+                    }
+                }
+            }
+            if self.display_thread_id {
+                write!(writer, "{:?}", thread.id())?;
+            }
+            writer.write_char('\n')?;
+        }
+
+        let bold = writer.bold();
+        let span = event
+            .parent()
+            .and_then(|id| ctx.span(id))
+            .or_else(|| ctx.lookup_current());
+
+        let scope = span.into_iter().flat_map(|span| span.scope());
+
+        for span in scope {
+            let meta = span.metadata();
+            if self.display_target {
+                write!(
+                    writer,
+                    "    {} {}::{}",
+                    dimmed.paint("in"),
+                    meta.target(),
+                    bold.paint(meta.name()),
+                )?;
+            } else {
+                write!(
+                    writer,
+                    "    {} {}",
+                    dimmed.paint("in"),
+                    bold.paint(meta.name()),
+                )?;
+            }
+
+            let ext = span.extensions();
+            let fields = &ext
+                .get::<FormattedFields<N>>()
+                .expect("Unable to find FormattedFields in extensions; this is a bug");
+            if !fields.is_empty() {
+                write!(writer, " {} {}", dimmed.paint("with"), fields)?;
+            }
+            writer.write_char('\n')?;
+        }
+
+        writer.write_char('\n')
+    }
+}
+
+impl<'writer> FormatFields<'writer> for Pretty {
+    fn format_fields<R: RecordFields>(&self, writer: Writer<'writer>, fields: R) -> fmt::Result {
+        let mut v = PrettyVisitor::new(writer, true);
+        fields.record(&mut v);
+        v.finish()
+    }
+
+    fn add_fields(
+        &self,
+        current: &'writer mut FormattedFields<Self>,
+        fields: &span::Record<'_>,
+    ) -> fmt::Result {
+        let empty = current.is_empty();
+        let writer = current.as_writer();
+        let mut v = PrettyVisitor::new(writer, empty);
+        fields.record(&mut v);
+        v.finish()
+    }
+}
+
+// === impl PrettyFields ===
+
+impl Default for PrettyFields {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
+impl PrettyFields {
+    /// Returns a new default [`PrettyFields`] implementation.
+    pub fn new() -> Self {
+        // By default, don't override the `Writer`'s ANSI colors
+        // configuration. We'll only do this if the user calls the
+        // deprecated `PrettyFields::with_ansi` method.
+        Self { ansi: None }
+    }
+
+    /// Enable ANSI encoding for formatted fields.
+    #[deprecated(
+        since = "0.3.3",
+        note = "Use `fmt::Subscriber::with_ansi` or `fmt::Layer::with_ansi` instead."
+    )]
+    pub fn with_ansi(self, ansi: bool) -> Self {
+        Self {
+            ansi: Some(ansi),
+            ..self
+        }
+    }
+}
+
+impl<'a> MakeVisitor<Writer<'a>> for PrettyFields {
+    type Visitor = PrettyVisitor<'a>;
+
+    #[inline]
+    fn make_visitor(&self, mut target: Writer<'a>) -> Self::Visitor {
+        if let Some(ansi) = self.ansi {
+            target = target.with_ansi(ansi);
+        }
+        PrettyVisitor::new(target, true)
+    }
+}
+
+// === impl PrettyVisitor ===
+
+impl<'a> PrettyVisitor<'a> {
+    /// Returns a new default visitor that formats to the provided `writer`.
+    ///
+    /// # Arguments
+    /// - `writer`: the writer to format to.
+    /// - `is_empty`: whether or not any fields have been previously written to
+    ///   that writer.
+    pub fn new(writer: Writer<'a>, is_empty: bool) -> Self {
+        Self {
+            writer,
+            is_empty,
+            style: Style::default(),
+            result: Ok(()),
+        }
+    }
+
+    pub(crate) fn with_style(self, style: Style) -> Self {
+        Self { style, ..self }
+    }
+
+    fn write_padded(&mut self, value: &impl fmt::Debug) {
+        let padding = if self.is_empty {
+            self.is_empty = false;
+            ""
+        } else {
+            ", "
+        };
+        self.result = write!(self.writer, "{}{:?}", padding, value);
+    }
+
+    fn bold(&self) -> Style {
+        if self.writer.has_ansi_escapes() {
+            self.style.bold()
+        } else {
+            Style::new()
+        }
+    }
+}
+
+impl field::Visit for PrettyVisitor<'_> {
+    fn record_str(&mut self, field: &Field, value: &str) {
+        if self.result.is_err() {
+            return;
+        }
+
+        if field.name() == "message" {
+            self.record_debug(field, &format_args!("{}", value))
+        } else {
+            self.record_debug(field, &value)
+        }
+    }
+
+    fn record_error(&mut self, field: &Field, value: &(dyn std::error::Error + 'static)) {
+        if let Some(source) = value.source() {
+            let bold = self.bold();
+            self.record_debug(
+                field,
+                &format_args!(
+                    "{}, {}{}.sources{}: {}",
+                    Escape(&format_args!("{}", value)),
+                    bold.prefix(),
+                    field,
+                    bold.infix(self.style),
+                    ErrorSourceList(source),
+                ),
+            )
+        } else {
+            self.record_debug(field, &Escape(&format_args!("{}", value)))
+        }
+    }
+
+    fn record_debug(&mut self, field: &Field, value: &dyn fmt::Debug) {
+        if self.result.is_err() {
+            return;
+        }
+        let bold = self.bold();
+        match field.name() {
+            "message" => {
+                // Escape ANSI characters to prevent malicious patterns (e.g., terminal injection attacks)
+                self.write_padded(&format_args!("{}{:?}", self.style.prefix(), Escape(value)))
+            },
+            // Skip fields that are actually log metadata that have already been handled
+            #[cfg(feature = "tracing-log")]
+            name if name.starts_with("log.") => self.result = Ok(()),
+            name if name.starts_with("r#") => self.write_padded(&format_args!(
+                "{}{}{}: {:?}",
+                bold.prefix(),
+                &name[2..],
+                bold.infix(self.style),
+                value
+            )),
+            name => self.write_padded(&format_args!(
+                "{}{}{}: {:?}",
+                bold.prefix(),
+                name,
+                bold.infix(self.style),
+                value
+            )),
+        };
+    }
+}
+
+impl VisitOutput<fmt::Result> for PrettyVisitor<'_> {
+    fn finish(mut self) -> fmt::Result {
+        write!(&mut self.writer, "{}", self.style.suffix())?;
+        self.result
+    }
+}
+
+impl VisitFmt for PrettyVisitor<'_> {
+    fn writer(&mut self) -> &mut dyn fmt::Write {
+        &mut self.writer
+    }
+}

vendor/tracing-subscriber/src/fmt/time/chrono_crate.rs

@@ -0,0 +1,174 @@
+use crate::fmt::format::Writer;
+use crate::fmt::time::FormatTime;
+
+use std::sync::Arc;
+
+/// Formats [local time]s and [UTC time]s with `FormatTime` implementations
+/// that use the [`chrono` crate].
+///
+/// [local time]: [`chrono::offset::Local`]
+/// [UTC time]: [`chrono::offset::Utc`]
+/// [`chrono` crate]: [`chrono`]
+
+/// Formats the current [local time] using a [formatter] from the [`chrono`] crate.
+///
+/// [local time]: chrono::Local::now()
+/// [formatter]: chrono::format
+#[cfg_attr(docsrs, doc(cfg(feature = "chrono")))]
+#[derive(Debug, Clone, Eq, PartialEq, Default)]
+pub struct ChronoLocal {
+    format: Arc<ChronoFmtType>,
+}
+
+impl ChronoLocal {
+    /// Format the time using the [`RFC 3339`] format
+    /// (a subset of [`ISO 8601`]).
+    ///
+    /// [`RFC 3339`]: https://tools.ietf.org/html/rfc3339
+    /// [`ISO 8601`]: https://en.wikipedia.org/wiki/ISO_8601
+    pub fn rfc_3339() -> Self {
+        Self {
+            format: Arc::new(ChronoFmtType::Rfc3339),
+        }
+    }
+
+    /// Format the time using the given format string.
+    ///
+    /// See [`chrono::format::strftime`] for details on the supported syntax.
+    pub fn new(format_string: String) -> Self {
+        Self {
+            format: Arc::new(ChronoFmtType::Custom(format_string)),
+        }
+    }
+}
+
+impl FormatTime for ChronoLocal {
+    fn format_time(&self, w: &mut Writer<'_>) -> alloc::fmt::Result {
+        let t = chrono::Local::now();
+        match self.format.as_ref() {
+            ChronoFmtType::Rfc3339 => {
+                use chrono::format::{Fixed, Item};
+                write!(
+                    w,
+                    "{}",
+                    t.format_with_items(core::iter::once(Item::Fixed(Fixed::RFC3339)))
+                )
+            }
+            ChronoFmtType::Custom(fmt) => {
+                write!(w, "{}", t.format(fmt))
+            }
+        }
+    }
+}
+
+/// Formats the current [UTC time] using a [formatter] from the [`chrono`] crate.
+///
+/// [UTC time]: chrono::Utc::now()
+/// [formatter]: chrono::format
+#[cfg_attr(docsrs, doc(cfg(feature = "chrono")))]
+#[derive(Debug, Clone, Eq, PartialEq, Default)]
+pub struct ChronoUtc {
+    format: Arc<ChronoFmtType>,
+}
+
+impl ChronoUtc {
+    /// Format the time using the [`RFC 3339`] format
+    /// (a subset of [`ISO 8601`]).
+    ///
+    /// [`RFC 3339`]: https://tools.ietf.org/html/rfc3339
+    /// [`ISO 8601`]: https://en.wikipedia.org/wiki/ISO_8601
+    pub fn rfc_3339() -> Self {
+        Self {
+            format: Arc::new(ChronoFmtType::Rfc3339),
+        }
+    }
+
+    /// Format the time using the given format string.
+    ///
+    /// See [`chrono::format::strftime`] for details on the supported syntax.
+    pub fn new(format_string: String) -> Self {
+        Self {
+            format: Arc::new(ChronoFmtType::Custom(format_string)),
+        }
+    }
+}
+
+impl FormatTime for ChronoUtc {
+    fn format_time(&self, w: &mut Writer<'_>) -> alloc::fmt::Result {
+        let t = chrono::Utc::now();
+        match self.format.as_ref() {
+            ChronoFmtType::Rfc3339 => w.write_str(&t.to_rfc3339()),
+            ChronoFmtType::Custom(fmt) => w.write_str(&format!("{}", t.format(fmt))),
+        }
+    }
+}
+
+/// The RFC 3339 format is used by default but a custom format string
+/// can be used. See [`chrono::format::strftime`]for details on
+/// the supported syntax.
+///
+/// [`chrono::format::strftime`]: https://docs.rs/chrono/0.4.9/chrono/format/strftime/index.html
+#[derive(Debug, Clone, Eq, PartialEq)]
+#[derive(Default)]
+enum ChronoFmtType {
+    /// Format according to the RFC 3339 convention.
+    #[default]
+    Rfc3339,
+    /// Format according to a custom format string.
+    Custom(String),
+}
+
+
+#[cfg(test)]
+mod tests {
+    use crate::fmt::format::Writer;
+    use crate::fmt::time::FormatTime;
+
+    use std::sync::Arc;
+
+    use super::ChronoFmtType;
+    use super::ChronoLocal;
+    use super::ChronoUtc;
+
+    #[test]
+    fn test_chrono_format_time_utc_default() {
+        let mut buf = String::new();
+        let mut dst: Writer<'_> = Writer::new(&mut buf);
+        assert!(FormatTime::format_time(&ChronoUtc::default(), &mut dst).is_ok());
+        // e.g. `buf` contains "2023-08-18T19:05:08.662499+00:00"
+        assert!(chrono::DateTime::parse_from_str(&buf, "%FT%H:%M:%S%.6f%z").is_ok());
+    }
+
+    #[test]
+    fn test_chrono_format_time_utc_custom() {
+        let fmt = ChronoUtc {
+            format: Arc::new(ChronoFmtType::Custom("%a %b %e %T %Y".to_owned())),
+        };
+        let mut buf = String::new();
+        let mut dst: Writer<'_> = Writer::new(&mut buf);
+        assert!(FormatTime::format_time(&fmt, &mut dst).is_ok());
+        // e.g. `buf` contains "Wed Aug 23 15:53:23 2023"
+        assert!(chrono::NaiveDateTime::parse_from_str(&buf, "%a %b %e %T %Y").is_ok());
+    }
+
+    #[test]
+    fn test_chrono_format_time_local_default() {
+        let mut buf = String::new();
+        let mut dst: Writer<'_> = Writer::new(&mut buf);
+        assert!(FormatTime::format_time(&ChronoLocal::default(), &mut dst).is_ok());
+        // e.g. `buf` contains "2023-08-18T14:59:08.662499-04:00".
+        assert!(chrono::DateTime::parse_from_str(&buf, "%FT%H:%M:%S%.6f%z").is_ok());
+    }
+
+    #[test]
+    fn test_chrono_format_time_local_custom() {
+        let fmt = ChronoLocal {
+            format: Arc::new(ChronoFmtType::Custom("%a %b %e %T %Y".to_owned())),
+        };
+        let mut buf = String::new();
+        let mut dst: Writer<'_> = Writer::new(&mut buf);
+        assert!(FormatTime::format_time(&fmt, &mut dst).is_ok());
+        // e.g. `buf` contains "Wed Aug 23 15:55:46 2023".
+        assert!(chrono::NaiveDateTime::parse_from_str(&buf, "%a %b %e %T %Y").is_ok());
+    }
+}

vendor/tracing-subscriber/src/fmt/time/datetime.rs

@@ -0,0 +1,410 @@
+// musl as a whole is licensed under the following standard MIT license:
+//
+// ----------------------------------------------------------------------
+// Copyright © 2005-2020 Rich Felker, et al.
+//
+// Permission is hereby granted, free of charge, to any person obtaining
+// a copy of this software and associated documentation files (the
+// "Software"), to deal in the Software without restriction, including
+// without limitation the rights to use, copy, modify, merge, publish,
+// distribute, sublicense, and/or sell copies of the Software, and to
+// permit persons to whom the Software is furnished to do so, subject to
+// the following conditions:
+//
+// The above copyright notice and this permission notice shall be
+// included in all copies or substantial portions of the Software.
+//
+// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+// EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+// MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+// IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
+// CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+// TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+// SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+// ----------------------------------------------------------------------
+//
+// Authors/contributors include:
+//
+// A. Wilcox
+// Ada Worcester
+// Alex Dowad
+// Alex Suykov
+// Alexander Monakov
+// Andre McCurdy
+// Andrew Kelley
+// Anthony G. Basile
+// Aric Belsito
+// Arvid Picciani
+// Bartosz Brachaczek
+// Benjamin Peterson
+// Bobby Bingham
+// Boris Brezillon
+// Brent Cook
+// Chris Spiegel
+// Clément Vasseur
+// Daniel Micay
+// Daniel Sabogal
+// Daurnimator
+// David Carlier
+// David Edelsohn
+// Denys Vlasenko
+// Dmitry Ivanov
+// Dmitry V. Levin
+// Drew DeVault
+// Emil Renner Berthing
+// Fangrui Song
+// Felix Fietkau
+// Felix Janda
+// Gianluca Anzolin
+// Hauke Mehrtens
+// He X
+// Hiltjo Posthuma
+// Isaac Dunham
+// Jaydeep Patil
+// Jens Gustedt
+// Jeremy Huntwork
+// Jo-Philipp Wich
+// Joakim Sindholt
+// John Spencer
+// Julien Ramseier
+// Justin Cormack
+// Kaarle Ritvanen
+// Khem Raj
+// Kylie McClain
+// Leah Neukirchen
+// Luca Barbato
+// Luka Perkov
+// M Farkas-Dyck (Strake)
+// Mahesh Bodapati
+// Markus Wichmann
+// Masanori Ogino
+// Michael Clark
+// Michael Forney
+// Mikhail Kremnyov
+// Natanael Copa
+// Nicholas J. Kain
+// orc
+// Pascal Cuoq
+// Patrick Oppenlander
+// Petr Hosek
+// Petr Skocik
+// Pierre Carrier
+// Reini Urban
+// Rich Felker
+// Richard Pennington
+// Ryan Fairfax
+// Samuel Holland
+// Segev Finer
+// Shiz
+// sin
+// Solar Designer
+// Stefan Kristiansson
+// Stefan O'Rear
+// Szabolcs Nagy
+// Timo Teräs
+// Trutz Behn
+// Valentin Ochs
+// Will Dietz
+// William Haddon
+// William Pitcock
+//
+// Portions of this software are derived from third-party works licensed
+// under terms compatible with the above MIT license:
+//
+// The TRE regular expression implementation (src/regex/reg* and
+// src/regex/tre*) is Copyright © 2001-2008 Ville Laurikari and licensed
+// under a 2-clause BSD license (license text in the source files). The
+// included version has been heavily modified by Rich Felker in 2012, in
+// the interests of size, simplicity, and namespace cleanliness.
+//
+// Much of the math library code (src/math/* and src/complex/*) is
+// Copyright © 1993,2004 Sun Microsystems or
+// Copyright © 2003-2011 David Schultz or
+// Copyright © 2003-2009 Steven G. Kargl or
+// Copyright © 2003-2009 Bruce D. Evans or
+// Copyright © 2008 Stephen L. Moshier or
+// Copyright © 2017-2018 Arm Limited
+// and labelled as such in comments in the individual source files. All
+// have been licensed under extremely permissive terms.
+//
+// The ARM memcpy code (src/string/arm/memcpy.S) is Copyright © 2008
+// The Android Open Source Project and is licensed under a two-clause BSD
+// license. It was taken from Bionic libc, used on Android.
+//
+// The AArch64 memcpy and memset code (src/string/aarch64/*) are
+// Copyright © 1999-2019, Arm Limited.
+//
+// The implementation of DES for crypt (src/crypt/crypt_des.c) is
+// Copyright © 1994 David Burren. It is licensed under a BSD license.
+//
+// The implementation of blowfish crypt (src/crypt/crypt_blowfish.c) was
+// originally written by Solar Designer and placed into the public
+// domain. The code also comes with a fallback permissive license for use
+// in jurisdictions that may not recognize the public domain.
+//
+// The smoothsort implementation (src/stdlib/qsort.c) is Copyright © 2011
+// Valentin Ochs and is licensed under an MIT-style license.
+//
+// The x86_64 port was written by Nicholas J. Kain and is licensed under
+// the standard MIT terms.
+//
+// The mips and microblaze ports were originally written by Richard
+// Pennington for use in the ellcc project. The original code was adapted
+// by Rich Felker for build system and code conventions during upstream
+// integration. It is licensed under the standard MIT terms.
+//
+// The mips64 port was contributed by Imagination Technologies and is
+// licensed under the standard MIT terms.
+//
+// The powerpc port was also originally written by Richard Pennington,
+// and later supplemented and integrated by John Spencer. It is licensed
+// under the standard MIT terms.
+//
+// All other files which have no copyright comments are original works
+// produced specifically for use as part of this library, written either
+// by Rich Felker, the main author of the library, or by one or more
+// contibutors listed above. Details on authorship of individual files
+// can be found in the git version control history of the project. The
+// omission of copyright and license comments in each file is in the
+// interest of source tree size.
+//
+// In addition, permission is hereby granted for all public header files
+// (include/* and arch/*/bits/*) and crt files intended to be linked into
+// applications (crt/*, ldso/dlstart.c, and arch/*/crt_arch.h) to omit
+// the copyright notice and permission notice otherwise required by the
+// license, and to use these files without any requirement of
+// attribution. These files include substantial contributions from:
+//
+// Bobby Bingham
+// John Spencer
+// Nicholas J. Kain
+// Rich Felker
+// Richard Pennington
+// Stefan Kristiansson
+// Szabolcs Nagy
+//
+// all of whom have explicitly granted such permission.
+//
+// This file previously contained text expressing a belief that most of
+// the files covered by the above exception were sufficiently trivial not
+// to be subject to copyright, resulting in confusion over whether it
+// negated the permissions granted in the license. In the spirit of
+// permissive licensing, and of not having licensing issues being an
+// obstacle to adoption, that text has been removed.
+
+
+use std::fmt;
+
+/// A date/time type which exists primarily to convert `SystemTime` timestamps into an ISO 8601
+/// formatted string.
+///
+/// Yes, this exists. Before you have a heart attack, understand that the meat of this is musl's
+/// [`__secs_to_tm`][1] converted to Rust via [c2rust][2] and then cleaned up by hand as part of
+/// the [kudu-rs project][3], [released under MIT][4].
+///
+/// [1] http://git.musl-libc.org/cgit/musl/tree/src/time/__secs_to_tm.c
+/// [2] https://c2rust.com/
+/// [3] https://github.com/danburkert/kudu-rs/blob/c9660067e5f4c1a54143f169b5eeb49446f82e54/src/timestamp.rs#L5-L18
+/// [4] https://github.com/tokio-rs/tracing/issues/1644#issuecomment-963888244
+///
+/// All existing `strftime`-like APIs I found were unable to handle the full range of timestamps representable
+/// by `SystemTime`, including `strftime` itself, since tm.tm_year is an int.
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub(crate) struct DateTime {
+    year: i64,
+    month: u8,
+    day: u8,
+    hour: u8,
+    minute: u8,
+    second: u8,
+    nanos: u32,
+}
+
+impl fmt::Display for DateTime {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        if self.year > 9999 {
+            write!(f, "+{}", self.year)?;
+        } else if self.year < 0 {
+            write!(f, "{:05}", self.year)?;
+        } else {
+            write!(f, "{:04}", self.year)?;
+        }
+
+        write!(
+            f,
+            "-{:02}-{:02}T{:02}:{:02}:{:02}.{:06}Z",
+            self.month,
+            self.day,
+            self.hour,
+            self.minute,
+            self.second,
+            self.nanos / 1_000
+        )
+    }
+}
+
+impl From<std::time::SystemTime> for DateTime {
+    fn from(timestamp: std::time::SystemTime) -> DateTime {
+        let (t, nanos) = match timestamp.duration_since(std::time::UNIX_EPOCH) {
+            Ok(duration) => {
+                debug_assert!(duration.as_secs() <= i64::MAX as u64);
+                (duration.as_secs() as i64, duration.subsec_nanos())
+            }
+            Err(error) => {
+                let duration = error.duration();
+                debug_assert!(duration.as_secs() <= i64::MAX as u64);
+                let (secs, nanos) = (duration.as_secs() as i64, duration.subsec_nanos());
+                if nanos == 0 {
+                    (-secs, 0)
+                } else {
+                    (-secs - 1, 1_000_000_000 - nanos)
+                }
+            }
+        };
+
+        // 2000-03-01 (mod 400 year, immediately after feb29
+        const LEAPOCH: i64 = 946_684_800 + 86400 * (31 + 29);
+        const DAYS_PER_400Y: i32 = 365 * 400 + 97;
+        const DAYS_PER_100Y: i32 = 365 * 100 + 24;
+        const DAYS_PER_4Y: i32 = 365 * 4 + 1;
+        static DAYS_IN_MONTH: [i8; 12] = [31, 30, 31, 30, 31, 31, 30, 31, 30, 31, 31, 29];
+
+        // Note(dcb): this bit is rearranged slightly to avoid integer overflow.
+        let mut days: i64 = (t / 86_400) - (LEAPOCH / 86_400);
+        let mut remsecs: i32 = (t % 86_400) as i32;
+        if remsecs < 0i32 {
+            remsecs += 86_400;
+            days -= 1
+        }
+
+        let mut qc_cycles: i32 = (days / i64::from(DAYS_PER_400Y)) as i32;
+        let mut remdays: i32 = (days % i64::from(DAYS_PER_400Y)) as i32;
+        if remdays < 0 {
+            remdays += DAYS_PER_400Y;
+            qc_cycles -= 1;
+        }
+
+        let mut c_cycles: i32 = remdays / DAYS_PER_100Y;
+        if c_cycles == 4 {
+            c_cycles -= 1;
+        }
+        remdays -= c_cycles * DAYS_PER_100Y;
+
+        let mut q_cycles: i32 = remdays / DAYS_PER_4Y;
+        if q_cycles == 25 {
+            q_cycles -= 1;
+        }
+        remdays -= q_cycles * DAYS_PER_4Y;
+
+        let mut remyears: i32 = remdays / 365;
+        if remyears == 4 {
+            remyears -= 1;
+        }
+        remdays -= remyears * 365;
+
+        let mut years: i64 = i64::from(remyears)
+            + 4 * i64::from(q_cycles)
+            + 100 * i64::from(c_cycles)
+            + 400 * i64::from(qc_cycles);
+
+        let mut months: i32 = 0;
+        while i32::from(DAYS_IN_MONTH[months as usize]) <= remdays {
+            remdays -= i32::from(DAYS_IN_MONTH[months as usize]);
+            months += 1
+        }
+
+        if months >= 10 {
+            months -= 12;
+            years += 1;
+        }
+
+        DateTime {
+            year: years + 2000,
+            month: (months + 3) as u8,
+            day: (remdays + 1) as u8,
+            hour: (remsecs / 3600) as u8,
+            minute: (remsecs / 60 % 60) as u8,
+            second: (remsecs % 60) as u8,
+            nanos,
+        }
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use i32;
+    use std::time::{Duration, UNIX_EPOCH};
+
+    use super::*;
+
+    #[test]
+    fn test_datetime() {
+        let case = |expected: &str, secs: i64, micros: u32| {
+            let timestamp = if secs >= 0 {
+                UNIX_EPOCH + Duration::new(secs as u64, micros * 1_000)
+            } else {
+                (UNIX_EPOCH - Duration::new(!secs as u64 + 1, 0)) + Duration::new(0, micros * 1_000)
+            };
+            assert_eq!(
+                expected,
+                format!("{}", DateTime::from(timestamp)),
+                "secs: {}, micros: {}",
+                secs,
+                micros
+            )
+        };
+
+        // Mostly generated with:
+        //  - date -jur <secs> +"%Y-%m-%dT%H:%M:%S.000000Z"
+        //  - http://unixtimestamp.50x.eu/
+
+        case("1970-01-01T00:00:00.000000Z", 0, 0);
+
+        case("1970-01-01T00:00:00.000001Z", 0, 1);
+        case("1970-01-01T00:00:00.500000Z", 0, 500_000);
+        case("1970-01-01T00:00:01.000001Z", 1, 1);
+        case("1970-01-01T00:01:01.000001Z", 60 + 1, 1);
+        case("1970-01-01T01:01:01.000001Z", 60 * 60 + 60 + 1, 1);
+        case(
+            "1970-01-02T01:01:01.000001Z",
+            24 * 60 * 60 + 60 * 60 + 60 + 1,
+            1,
+        );
+
+        case("1969-12-31T23:59:59.000000Z", -1, 0);
+        case("1969-12-31T23:59:59.000001Z", -1, 1);
+        case("1969-12-31T23:59:59.500000Z", -1, 500_000);
+        case("1969-12-31T23:58:59.000001Z", -60 - 1, 1);
+        case("1969-12-31T22:58:59.000001Z", -60 * 60 - 60 - 1, 1);
+        case(
+            "1969-12-30T22:58:59.000001Z",
+            -24 * 60 * 60 - 60 * 60 - 60 - 1,
+            1,
+        );
+
+        case("2038-01-19T03:14:07.000000Z", i32::MAX as i64, 0);
+        case("2038-01-19T03:14:08.000000Z", i32::MAX as i64 + 1, 0);
+        case("1901-12-13T20:45:52.000000Z", i32::MIN as i64, 0);
+        case("1901-12-13T20:45:51.000000Z", i32::MIN as i64 - 1, 0);
+
+        // Skipping these tests on windows as std::time::SystemTime range is low
+        // on Windows compared with that of Unix which can cause the following
+        // high date value tests to panic
+        #[cfg(not(target_os = "windows"))]
+        {
+            case("+292277026596-12-04T15:30:07.000000Z", i64::MAX, 0);
+            case("+292277026596-12-04T15:30:06.000000Z", i64::MAX - 1, 0);
+            case("-292277022657-01-27T08:29:53.000000Z", i64::MIN + 1, 0);
+        }
+
+        case("1900-01-01T00:00:00.000000Z", -2208988800, 0);
+        case("1899-12-31T23:59:59.000000Z", -2208988801, 0);
+        case("0000-01-01T00:00:00.000000Z", -62167219200, 0);
+        case("-0001-12-31T23:59:59.000000Z", -62167219201, 0);
+
+        case("1234-05-06T07:08:09.000000Z", -23215049511, 0);
+        case("-1234-05-06T07:08:09.000000Z", -101097651111, 0);
+        case("2345-06-07T08:09:01.000000Z", 11847456541, 0);
+        case("-2345-06-07T08:09:01.000000Z", -136154620259, 0);
+    }
+}

vendor/tracing-subscriber/src/fmt/time/mod.rs

@@ -0,0 +1,151 @@
+//! Formatters for event timestamps.
+use crate::fmt::format::Writer;
+use std::fmt;
+use std::time::Instant;
+
+mod datetime;
+
+#[cfg(feature = "time")]
+mod time_crate;
+
+#[cfg(feature = "time")]
+#[cfg_attr(docsrs, doc(cfg(feature = "time")))]
+pub use time_crate::UtcTime;
+
+#[cfg(feature = "local-time")]
+#[cfg_attr(docsrs, doc(cfg(all(unsound_local_offset, feature = "local-time"))))]
+pub use time_crate::LocalTime;
+
+#[cfg(feature = "time")]
+#[cfg_attr(docsrs, doc(cfg(feature = "time")))]
+pub use time_crate::OffsetTime;
+
+/// [`chrono`]-based implementation for [`FormatTime`].
+#[cfg(feature = "chrono")]
+mod chrono_crate;
+
+#[cfg(feature = "chrono")]
+#[cfg_attr(docsrs, doc(cfg(feature = "chrono")))]
+pub use chrono_crate::ChronoLocal;
+
+#[cfg(feature = "chrono")]
+#[cfg_attr(docsrs, doc(cfg(feature = "chrono")))]
+pub use chrono_crate::ChronoUtc;
+
+/// A type that can measure and format the current time.
+///
+/// This trait is used by `Format` to include a timestamp with each `Event` when it is logged.
+///
+/// Notable default implementations of this trait are `SystemTime` and `()`. The former prints the
+/// current time as reported by `std::time::SystemTime`, and the latter does not print the current
+/// time at all. `FormatTime` is also automatically implemented for any function pointer with the
+/// appropriate signature.
+///
+/// The full list of provided implementations can be found in [`time`].
+///
+/// [`time`]: self
+pub trait FormatTime {
+    /// Measure and write out the current time.
+    ///
+    /// When `format_time` is called, implementors should get the current time using their desired
+    /// mechanism, and write it out to the given `fmt::Write`. Implementors must insert a trailing
+    /// space themselves if they wish to separate the time from subsequent log message text.
+    fn format_time(&self, w: &mut Writer<'_>) -> fmt::Result;
+}
+
+/// Returns a new `SystemTime` timestamp provider.
+///
+/// This can then be configured further to determine how timestamps should be
+/// configured.
+///
+/// This is equivalent to calling
+/// ```rust
+/// # fn timer() -> tracing_subscriber::fmt::time::SystemTime {
+/// tracing_subscriber::fmt::time::SystemTime::default()
+/// # }
+/// ```
+pub fn time() -> SystemTime {
+    SystemTime
+}
+
+/// Returns a new `Uptime` timestamp provider.
+///
+/// With this timer, timestamps will be formatted with the amount of time
+/// elapsed since the timestamp provider was constructed.
+///
+/// This can then be configured further to determine how timestamps should be
+/// configured.
+///
+/// This is equivalent to calling
+/// ```rust
+/// # fn timer() -> tracing_subscriber::fmt::time::Uptime {
+/// tracing_subscriber::fmt::time::Uptime::default()
+/// # }
+/// ```
+pub fn uptime() -> Uptime {
+    Uptime::default()
+}
+
+impl<F> FormatTime for &F
+where
+    F: FormatTime,
+{
+    fn format_time(&self, w: &mut Writer<'_>) -> fmt::Result {
+        (*self).format_time(w)
+    }
+}
+
+impl FormatTime for () {
+    fn format_time(&self, _: &mut Writer<'_>) -> fmt::Result {
+        Ok(())
+    }
+}
+
+impl FormatTime for fn(&mut Writer<'_>) -> fmt::Result {
+    fn format_time(&self, w: &mut Writer<'_>) -> fmt::Result {
+        (*self)(w)
+    }
+}
+
+/// Retrieve and print the current wall-clock time.
+#[derive(Debug, Clone, Copy, Eq, PartialEq, Default)]
+pub struct SystemTime;
+
+/// Retrieve and print the relative elapsed wall-clock time since an epoch.
+///
+/// The `Default` implementation for `Uptime` makes the epoch the current time.
+#[derive(Debug, Clone, Copy, Eq, PartialEq)]
+pub struct Uptime {
+    epoch: Instant,
+}
+
+impl Default for Uptime {
+    fn default() -> Self {
+        Uptime {
+            epoch: Instant::now(),
+        }
+    }
+}
+
+impl From<Instant> for Uptime {
+    fn from(epoch: Instant) -> Self {
+        Uptime { epoch }
+    }
+}
+
+impl FormatTime for SystemTime {
+    fn format_time(&self, w: &mut Writer<'_>) -> fmt::Result {
+        write!(
+            w,
+            "{}",
+            datetime::DateTime::from(std::time::SystemTime::now())
+        )
+    }
+}
+
+impl FormatTime for Uptime {
+    fn format_time(&self, w: &mut Writer<'_>) -> fmt::Result {
+        let e = self.epoch.elapsed();
+        write!(w, "{:4}.{:09}s", e.as_secs(), e.subsec_nanos())
+    }
+}

vendor/tracing-subscriber/src/fmt/time/time_crate.rs

@@ -0,0 +1,470 @@
+use crate::fmt::{format::Writer, time::FormatTime, writer::WriteAdaptor};
+use std::fmt;
+use time::{format_description::well_known, formatting::Formattable, OffsetDateTime, UtcOffset};
+
+/// Formats the current [local time] using a [formatter] from the [`time` crate].
+///
+/// To format the current [UTC time] instead, use the [`UtcTime`] type.
+///
+/// <div class="example-wrap" style="display:inline-block">
+/// <pre class="compile_fail" style="white-space:normal;font:inherit;">
+///     <strong>Warning</strong>: The <a href = "https://docs.rs/time/0.3/time/"><code>time</code>
+///     crate</a> must be compiled with <code>--cfg unsound_local_offset</code> in order to use
+///     local timestamps. When this cfg is not enabled, local timestamps cannot be recorded, and
+///     events will be logged without timestamps.
+///
+///    Alternatively, [`OffsetTime`] can log with a local offset if it is initialized early.
+///
+///    See the <a href="https://docs.rs/time/0.3.4/time/#feature-flags"><code>time</code>
+///    documentation</a> for more details.
+/// </pre></div>
+///
+/// [local time]: time::OffsetDateTime::now_local
+/// [UTC time]:     time::OffsetDateTime::now_utc
+/// [formatter]:    time::formatting::Formattable
+/// [`time` crate]: time
+#[derive(Clone, Debug)]
+#[cfg_attr(
+    docsrs,
+    doc(cfg(all(unsound_local_offset, feature = "time", feature = "local-time")))
+)]
+#[cfg(feature = "local-time")]
+pub struct LocalTime<F> {
+    format: F,
+}
+
+/// Formats the current [UTC time] using a [formatter] from the [`time` crate].
+///
+/// To format the current [local time] instead, use the [`LocalTime`] type.
+///
+/// [local time]: time::OffsetDateTime::now_local
+/// [UTC time]:     time::OffsetDateTime::now_utc
+/// [formatter]:    time::formatting::Formattable
+/// [`time` crate]: time
+#[cfg_attr(docsrs, doc(cfg(feature = "time")))]
+#[derive(Clone, Debug)]
+pub struct UtcTime<F> {
+    format: F,
+}
+
+/// Formats the current time using a fixed offset and a [formatter] from the [`time` crate].
+///
+/// This is typically used as an alternative to [`LocalTime`]. `LocalTime` determines the offset
+/// every time it formats a message, which may be unsound or fail. With `OffsetTime`, the offset is
+/// determined once. This makes it possible to do so while the program is still single-threaded and
+/// handle any errors. However, this also means the offset cannot change while the program is
+/// running (the offset will not change across DST changes).
+///
+/// [formatter]: time::formatting::Formattable
+/// [`time` crate]: time
+#[derive(Clone, Debug)]
+#[cfg_attr(docsrs, doc(cfg(feature = "time")))]
+pub struct OffsetTime<F> {
+    offset: time::UtcOffset,
+    format: F,
+}
+
+// === impl LocalTime ===
+
+#[cfg(feature = "local-time")]
+impl LocalTime<well_known::Rfc3339> {
+    /// Returns a formatter that formats the current [local time] in the
+    /// [RFC 3339] format (a subset of the [ISO 8601] timestamp format).
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// use tracing_subscriber::fmt::{self, time};
+    ///
+    /// let subscriber = tracing_subscriber::fmt()
+    ///     .with_timer(time::LocalTime::rfc_3339());
+    /// # drop(subscriber);
+    /// ```
+    ///
+    /// [local time]: time::OffsetDateTime::now_local
+    /// [RFC 3339]: https://datatracker.ietf.org/doc/html/rfc3339
+    /// [ISO 8601]: https://en.wikipedia.org/wiki/ISO_8601
+    pub fn rfc_3339() -> Self {
+        Self::new(well_known::Rfc3339)
+    }
+}
+
+#[cfg(feature = "local-time")]
+impl<F: Formattable> LocalTime<F> {
+    /// Returns a formatter that formats the current [local time] using the
+    /// [`time` crate] with the provided provided format. The format may be any
+    /// type that implements the [`Formattable`] trait.
+    ///
+    ///
+    /// <div class="example-wrap" style="display:inline-block">
+    /// <pre class="compile_fail" style="white-space:normal;font:inherit;">
+    ///     <strong>Warning</strong>: The <a href = "https://docs.rs/time/0.3/time/">
+    ///     <code>time</code> crate</a> must be compiled with <code>--cfg
+    ///     unsound_local_offset</code> in order to use local timestamps. When this
+    ///     cfg is not enabled, local timestamps cannot be recorded, and
+    ///     events will be logged without timestamps.
+    ///
+    ///    See the <a href="https://docs.rs/time/0.3.4/time/#feature-flags">
+    ///    <code>time</code> documentation</a> for more details.
+    /// </pre></div>
+    ///
+    /// Typically, the format will be a format description string, or one of the
+    /// `time` crate's [well-known formats].
+    ///
+    /// If the format description is statically known, then the
+    /// [`format_description!`] macro should be used. This is identical to the
+    /// [`time::format_description::parse`] method, but runs at compile-time,
+    /// throwing an error if the format description is invalid. If the desired format
+    /// is not known statically (e.g., a user is providing a format string), then the
+    /// [`time::format_description::parse`] method should be used. Note that this
+    /// method is fallible.
+    ///
+    /// See the [`time` book] for details on the format description syntax.
+    ///
+    /// # Examples
+    ///
+    /// Using the [`format_description!`] macro:
+    ///
+    /// ```
+    /// use tracing_subscriber::fmt::{self, time::LocalTime};
+    /// use time::macros::format_description;
+    ///
+    /// let timer = LocalTime::new(format_description!("[hour]:[minute]:[second]"));
+    /// let subscriber = tracing_subscriber::fmt()
+    ///     .with_timer(timer);
+    /// # drop(subscriber);
+    /// ```
+    ///
+    /// Using [`time::format_description::parse`]:
+    ///
+    /// ```
+    /// use tracing_subscriber::fmt::{self, time::LocalTime};
+    ///
+    /// let time_format = time::format_description::parse("[hour]:[minute]:[second]")
+    ///     .expect("format string should be valid!");
+    /// let timer = LocalTime::new(time_format);
+    /// let subscriber = tracing_subscriber::fmt()
+    ///     .with_timer(timer);
+    /// # drop(subscriber);
+    /// ```
+    ///
+    /// Using the [`format_description!`] macro requires enabling the `time`
+    /// crate's "macros" feature flag.
+    ///
+    /// Using a [well-known format][well-known formats] (this is equivalent to
+    /// [`LocalTime::rfc_3339`]):
+    ///
+    /// ```
+    /// use tracing_subscriber::fmt::{self, time::LocalTime};
+    ///
+    /// let timer = LocalTime::new(time::format_description::well_known::Rfc3339);
+    /// let subscriber = tracing_subscriber::fmt()
+    ///     .with_timer(timer);
+    /// # drop(subscriber);
+    /// ```
+    ///
+    /// [local time]: time::OffsetDateTime::now_local()
+    /// [`time` crate]: time
+    /// [`Formattable`]: time::formatting::Formattable
+    /// [well-known formats]: time::format_description::well_known
+    /// [`format_description!`]: https://docs.rs/time/0.3/time/macros/macro.format_description.html
+    /// [`time::format_description::parse`]: time::format_description::parse()
+    /// [`time` book]: https://time-rs.github.io/book/api/format-description.html
+    pub fn new(format: F) -> Self {
+        Self { format }
+    }
+}
+
+#[cfg(feature = "local-time")]
+impl<F> FormatTime for LocalTime<F>
+where
+    F: Formattable,
+{
+    fn format_time(&self, w: &mut Writer<'_>) -> fmt::Result {
+        let now = OffsetDateTime::now_local().map_err(|_| fmt::Error)?;
+        format_datetime(now, w, &self.format)
+    }
+}
+
+#[cfg(feature = "local-time")]
+impl<F> Default for LocalTime<F>
+where
+    F: Formattable + Default,
+{
+    fn default() -> Self {
+        Self::new(F::default())
+    }
+}
+
+// === impl UtcTime ===
+
+impl UtcTime<well_known::Rfc3339> {
+    /// Returns a formatter that formats the current [UTC time] in the
+    /// [RFC 3339] format, which is a subset of the [ISO 8601] timestamp format.
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// use tracing_subscriber::fmt::{self, time};
+    ///
+    /// let subscriber = tracing_subscriber::fmt()
+    ///     .with_timer(time::UtcTime::rfc_3339());
+    /// # drop(subscriber);
+    /// ```
+    ///
+    /// [local time]: time::OffsetDateTime::now_utc
+    /// [RFC 3339]: https://datatracker.ietf.org/doc/html/rfc3339
+    /// [ISO 8601]: https://en.wikipedia.org/wiki/ISO_8601
+    pub fn rfc_3339() -> Self {
+        Self::new(well_known::Rfc3339)
+    }
+}
+
+impl<F: Formattable> UtcTime<F> {
+    /// Returns a formatter that formats the current [UTC time] using the
+    /// [`time` crate], with the provided provided format. The format may be any
+    /// type that implements the [`Formattable`] trait.
+    ///
+    /// Typically, the format will be a format description string, or one of the
+    /// `time` crate's [well-known formats].
+    ///
+    /// If the format description is statically known, then the
+    /// [`format_description!`] macro should be used. This is identical to the
+    /// [`time::format_description::parse`] method, but runs at compile-time,
+    /// failing  an error if the format description is invalid. If the desired format
+    /// is not known statically (e.g., a user is providing a format string), then the
+    /// [`time::format_description::parse`] method should be used. Note that this
+    /// method is fallible.
+    ///
+    /// See the [`time` book] for details on the format description syntax.
+    ///
+    /// # Examples
+    ///
+    /// Using the [`format_description!`] macro:
+    ///
+    /// ```
+    /// use tracing_subscriber::fmt::{self, time::UtcTime};
+    /// use time::macros::format_description;
+    ///
+    /// let timer = UtcTime::new(format_description!("[hour]:[minute]:[second]"));
+    /// let subscriber = tracing_subscriber::fmt()
+    ///     .with_timer(timer);
+    /// # drop(subscriber);
+    /// ```
+    ///
+    /// Using the [`format_description!`] macro requires enabling the `time`
+    /// crate's "macros" feature flag.
+    ///
+    /// Using [`time::format_description::parse`]:
+    ///
+    /// ```
+    /// use tracing_subscriber::fmt::{self, time::UtcTime};
+    ///
+    /// let time_format = time::format_description::parse("[hour]:[minute]:[second]")
+    ///     .expect("format string should be valid!");
+    /// let timer = UtcTime::new(time_format);
+    /// let subscriber = tracing_subscriber::fmt()
+    ///     .with_timer(timer);
+    /// # drop(subscriber);
+    /// ```
+    ///
+    /// Using a [well-known format][well-known formats] (this is equivalent to
+    /// [`UtcTime::rfc_3339`]):
+    ///
+    /// ```
+    /// use tracing_subscriber::fmt::{self, time::UtcTime};
+    ///
+    /// let timer = UtcTime::new(time::format_description::well_known::Rfc3339);
+    /// let subscriber = tracing_subscriber::fmt()
+    ///     .with_timer(timer);
+    /// # drop(subscriber);
+    /// ```
+    ///
+    /// [UTC time]: time::OffsetDateTime::now_utc()
+    /// [`time` crate]: time
+    /// [`Formattable`]: time::formatting::Formattable
+    /// [well-known formats]: time::format_description::well_known
+    /// [`format_description!`]: https://docs.rs/time/0.3/time/macros/macro.format_description.html
+    /// [`time::format_description::parse`]: time::format_description::parse
+    /// [`time` book]: https://time-rs.github.io/book/api/format-description.html
+    pub fn new(format: F) -> Self {
+        Self { format }
+    }
+}
+
+impl<F> FormatTime for UtcTime<F>
+where
+    F: Formattable,
+{
+    fn format_time(&self, w: &mut Writer<'_>) -> fmt::Result {
+        format_datetime(OffsetDateTime::now_utc(), w, &self.format)
+    }
+}
+
+impl<F> Default for UtcTime<F>
+where
+    F: Formattable + Default,
+{
+    fn default() -> Self {
+        Self::new(F::default())
+    }
+}
+
+// === impl OffsetTime ===
+
+#[cfg(feature = "local-time")]
+impl OffsetTime<well_known::Rfc3339> {
+    /// Returns a formatter that formats the current time using the [local time offset] in the [RFC
+    /// 3339] format (a subset of the [ISO 8601] timestamp format).
+    ///
+    /// Returns an error if the local time offset cannot be determined. This typically occurs in
+    /// multithreaded programs. To avoid this problem, initialize `OffsetTime` before forking
+    /// threads. When using Tokio, this means initializing `OffsetTime` before the Tokio runtime.
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// use tracing_subscriber::fmt::{self, time};
+    ///
+    /// let subscriber = tracing_subscriber::fmt()
+    ///     .with_timer(time::OffsetTime::local_rfc_3339().expect("could not get local offset!"));
+    /// # drop(subscriber);
+    /// ```
+    ///
+    /// Using `OffsetTime` with Tokio:
+    ///
+    /// ```
+    /// use tracing_subscriber::fmt::time::OffsetTime;
+    ///
+    /// #[tokio::main]
+    /// async fn run() {
+    ///     tracing::info!("runtime initialized");
+    ///
+    ///     // At this point the Tokio runtime is initialized, and we can use both Tokio and Tracing
+    ///     // normally.
+    /// }
+    ///
+    /// fn main() {
+    ///     // Because we need to get the local offset before Tokio spawns any threads, our `main`
+    ///     // function cannot use `tokio::main`.
+    ///     tracing_subscriber::fmt()
+    ///         .with_timer(OffsetTime::local_rfc_3339().expect("could not get local time offset"))
+    ///         .init();
+    ///
+    ///     // Even though `run` is written as an `async fn`, because we used `tokio::main` on it
+    ///     // we can call it as a synchronous function.
+    ///     run();
+    /// }
+    /// ```
+    ///
+    /// [local time offset]: time::UtcOffset::current_local_offset
+    /// [RFC 3339]: https://datatracker.ietf.org/doc/html/rfc3339
+    /// [ISO 8601]: https://en.wikipedia.org/wiki/ISO_8601
+    pub fn local_rfc_3339() -> Result<Self, time::error::IndeterminateOffset> {
+        Ok(Self::new(
+            UtcOffset::current_local_offset()?,
+            well_known::Rfc3339,
+        ))
+    }
+}
+
+impl<F: time::formatting::Formattable> OffsetTime<F> {
+    /// Returns a formatter that formats the current time using the [`time` crate] with the provided
+    /// provided format and [timezone offset]. The format may be any type that implements the
+    /// [`Formattable`] trait.
+    ///
+    ///
+    /// Typically, the offset will be the [local offset], and format will be a format description
+    /// string, or one of the `time` crate's [well-known formats].
+    ///
+    /// If the format description is statically known, then the
+    /// [`format_description!`] macro should be used. This is identical to the
+    /// [`time::format_description::parse`] method, but runs at compile-time,
+    /// throwing an error if the format description is invalid. If the desired format
+    /// is not known statically (e.g., a user is providing a format string), then the
+    /// [`time::format_description::parse`] method should be used. Note that this
+    /// method is fallible.
+    ///
+    /// See the [`time` book] for details on the format description syntax.
+    ///
+    /// # Examples
+    ///
+    /// Using the [`format_description!`] macro:
+    ///
+    /// ```
+    /// use tracing_subscriber::fmt::{self, time::OffsetTime};
+    /// use time::macros::format_description;
+    /// use time::UtcOffset;
+    ///
+    /// let offset = UtcOffset::current_local_offset().expect("should get local offset!");
+    /// let timer = OffsetTime::new(offset, format_description!("[hour]:[minute]:[second]"));
+    /// let subscriber = tracing_subscriber::fmt()
+    ///     .with_timer(timer);
+    /// # drop(subscriber);
+    /// ```
+    ///
+    /// Using [`time::format_description::parse`]:
+    ///
+    /// ```
+    /// use tracing_subscriber::fmt::{self, time::OffsetTime};
+    /// use time::UtcOffset;
+    ///
+    /// let offset = UtcOffset::current_local_offset().expect("should get local offset!");
+    /// let time_format = time::format_description::parse("[hour]:[minute]:[second]")
+    ///     .expect("format string should be valid!");
+    /// let timer = OffsetTime::new(offset, time_format);
+    /// let subscriber = tracing_subscriber::fmt()
+    ///     .with_timer(timer);
+    /// # drop(subscriber);
+    /// ```
+    ///
+    /// Using the [`format_description!`] macro requires enabling the `time`
+    /// crate's "macros" feature flag.
+    ///
+    /// Using a [well-known format][well-known formats] (this is equivalent to
+    /// [`OffsetTime::local_rfc_3339`]):
+    ///
+    /// ```
+    /// use tracing_subscriber::fmt::{self, time::OffsetTime};
+    /// use time::UtcOffset;
+    ///
+    /// let offset = UtcOffset::current_local_offset().expect("should get local offset!");
+    /// let timer = OffsetTime::new(offset, time::format_description::well_known::Rfc3339);
+    /// let subscriber = tracing_subscriber::fmt()
+    ///     .with_timer(timer);
+    /// # drop(subscriber);
+    /// ```
+    ///
+    /// [`time` crate]: time
+    /// [timezone offset]: time::UtcOffset
+    /// [`Formattable`]: time::formatting::Formattable
+    /// [local offset]: time::UtcOffset::current_local_offset()
+    /// [well-known formats]: time::format_description::well_known
+    /// [`format_description!`]: https://docs.rs/time/0.3/time/macros/macro.format_description.html
+    /// [`time::format_description::parse`]: time::format_description::parse
+    /// [`time` book]: https://time-rs.github.io/book/api/format-description.html
+    pub fn new(offset: time::UtcOffset, format: F) -> Self {
+        Self { offset, format }
+    }
+}
+
+impl<F> FormatTime for OffsetTime<F>
+where
+    F: time::formatting::Formattable,
+{
+    fn format_time(&self, w: &mut Writer<'_>) -> fmt::Result {
+        let now = OffsetDateTime::now_utc().to_offset(self.offset);
+        format_datetime(now, w, &self.format)
+    }
+}
+
+fn format_datetime(
+    now: OffsetDateTime,
+    into: &mut Writer<'_>,
+    fmt: &impl Formattable,
+) -> fmt::Result {
+    let mut into = WriteAdaptor::new(into);
+    now.format_into(&mut into, fmt)
+        .map_err(|_| fmt::Error)
+        .map(|_| ())
+}

vendor/tracing-subscriber/src/layer/context.rs

@@ -0,0 +1,433 @@
+use tracing_core::{metadata::Metadata, span, subscriber::Subscriber, Event};
+
+use crate::registry::{self, LookupSpan, SpanRef};
+
+#[cfg(all(feature = "registry", feature = "std"))]
+use crate::{filter::FilterId, registry::Registry};
+/// Represents information about the current context provided to [`Layer`]s by the
+/// wrapped [`Subscriber`].
+///
+/// To access [stored data] keyed by a span ID, implementors of the `Layer`
+/// trait should ensure that the `Subscriber` type parameter is *also* bound by the
+/// [`LookupSpan`]:
+///
+/// ```rust
+/// use tracing::Subscriber;
+/// use tracing_subscriber::{Layer, registry::LookupSpan};
+///
+/// pub struct MyLayer;
+///
+/// impl<S> Layer<S> for MyLayer
+/// where
+///     S: Subscriber + for<'a> LookupSpan<'a>,
+/// {
+///     // ...
+/// }
+/// ```
+///
+/// [`Layer`]: super::Layer
+/// [`Subscriber`]: tracing_core::Subscriber
+/// [stored data]: crate::registry::SpanRef
+/// [`LookupSpan`]: crate::registry::LookupSpan
+#[derive(Debug)]
+pub struct Context<'a, S> {
+    subscriber: Option<&'a S>,
+    /// The bitmask of all [`Filtered`] layers that currently apply in this
+    /// context. If there is only a single [`Filtered`] wrapping the layer that
+    /// produced this context, then this is that filter's ID. Otherwise, if we
+    /// are in a nested tree with multiple filters, this is produced by
+    /// [`and`]-ing together the [`FilterId`]s of each of the filters that wrap
+    /// the current layer.
+    ///
+    /// [`Filtered`]: crate::filter::Filtered
+    /// [`FilterId`]: crate::filter::FilterId
+    /// [`and`]: crate::filter::FilterId::and
+    #[cfg(all(feature = "registry", feature = "std"))]
+    filter: FilterId,
+}
+
+// === impl Context ===
+
+impl<'a, S> Context<'a, S>
+where
+    S: Subscriber,
+{
+    pub(super) fn new(subscriber: &'a S) -> Self {
+        Self {
+            subscriber: Some(subscriber),
+
+            #[cfg(feature = "registry")]
+            filter: FilterId::none(),
+        }
+    }
+
+    /// Returns the wrapped subscriber's view of the current span.
+    #[inline]
+    pub fn current_span(&self) -> span::Current {
+        self.subscriber
+            .map(Subscriber::current_span)
+            // TODO: this would be more correct as "unknown", so perhaps
+            // `tracing-core` should make `Current::unknown()` public?
+            .unwrap_or_else(span::Current::none)
+    }
+
+    /// Returns whether the wrapped subscriber would enable the current span.
+    #[inline]
+    pub fn enabled(&self, metadata: &Metadata<'_>) -> bool {
+        self.subscriber
+            .map(|subscriber| subscriber.enabled(metadata))
+            // If this context is `None`, we are registering a callsite, so
+            // return `true` so that the layer does not incorrectly assume that
+            // the inner subscriber has disabled this metadata.
+            // TODO(eliza): would it be more correct for this to return an `Option`?
+            .unwrap_or(true)
+    }
+
+    /// Records the provided `event` with the wrapped subscriber.
+    ///
+    /// # Notes
+    ///
+    /// - The subscriber is free to expect that the event's callsite has been
+    ///   [registered][register], and may panic or fail to observe the event if this is
+    ///   not the case. The `tracing` crate's macros ensure that all events are
+    ///   registered, but if the event is constructed through other means, the
+    ///   user is responsible for ensuring that [`register_callsite`][register]
+    ///   has been called prior to calling this method.
+    /// - This does _not_ call [`enabled`] on the inner subscriber. If the
+    ///   caller wishes to apply the wrapped subscriber's filter before choosing
+    ///   whether to record the event, it may first call [`Context::enabled`] to
+    ///   check whether the event would be enabled. This allows `Layer`s to
+    ///   elide constructing the event if it would not be recorded.
+    ///
+    /// [register]: tracing_core::subscriber::Subscriber::register_callsite()
+    /// [`enabled`]: tracing_core::subscriber::Subscriber::enabled()
+    /// [`Context::enabled`]: Context::enabled()
+    #[inline]
+    pub fn event(&self, event: &Event<'_>) {
+        if let Some(subscriber) = self.subscriber {
+            subscriber.event(event);
+        }
+    }
+
+    /// Returns a [`SpanRef`] for the parent span of the given [`Event`], if
+    /// it has a parent.
+    ///
+    /// If the event has an explicitly overridden parent, this method returns
+    /// a reference to that span. If the event's parent is the current span,
+    /// this returns a reference to the current span, if there is one. If this
+    /// returns `None`, then either the event's parent was explicitly set to
+    /// `None`, or the event's parent was defined contextually, but no span
+    /// is currently entered.
+    ///
+    /// Compared to [`Context::current_span`] and [`Context::lookup_current`],
+    /// this respects overrides provided by the [`Event`].
+    ///
+    /// Compared to [`Event::parent`], this automatically falls back to the contextual
+    /// span, if required.
+    ///
+    /// ```rust
+    /// use tracing::{Event, Subscriber};
+    /// use tracing_subscriber::{
+    ///     layer::{Context, Layer},
+    ///     prelude::*,
+    ///     registry::LookupSpan,
+    /// };
+    ///
+    /// struct PrintingLayer;
+    /// impl<S> Layer<S> for PrintingLayer
+    /// where
+    ///     S: Subscriber + for<'lookup> LookupSpan<'lookup>,
+    /// {
+    ///     fn on_event(&self, event: &Event, ctx: Context<S>) {
+    ///         let span = ctx.event_span(event);
+    ///         println!("Event in span: {:?}", span.map(|s| s.name()));
+    ///     }
+    /// }
+    ///
+    /// tracing::subscriber::with_default(tracing_subscriber::registry().with(PrintingLayer), || {
+    ///     tracing::info!("no span");
+    ///     // Prints: Event in span: None
+    ///
+    ///     let span = tracing::info_span!("span");
+    ///     tracing::info!(parent: &span, "explicitly specified");
+    ///     // Prints: Event in span: Some("span")
+    ///
+    ///     let _guard = span.enter();
+    ///     tracing::info!("contextual span");
+    ///     // Prints: Event in span: Some("span")
+    /// });
+    /// ```
+    ///
+    /// <pre class="ignore" style="white-space:normal;font:inherit;">
+    ///     <strong>Note</strong>: This requires the wrapped subscriber to
+    ///     implement the <a href="../registry/trait.LookupSpan.html"><code>
+    ///     LookupSpan</code></a> trait. See the documentation on
+    ///     <a href="./struct.Context.html"><code>Context</code>'s
+    ///     declaration</a> for details.
+    /// </pre>
+    #[inline]
+    pub fn event_span(&self, event: &Event<'_>) -> Option<SpanRef<'_, S>>
+    where
+        S: for<'lookup> LookupSpan<'lookup>,
+    {
+        if event.is_root() {
+            None
+        } else if event.is_contextual() {
+            self.lookup_current()
+        } else {
+            // TODO(eliza): this should handle parent IDs
+            event.parent().and_then(|id| self.span(id))
+        }
+    }
+
+    /// Returns metadata for the span with the given `id`, if it exists.
+    ///
+    /// If this returns `None`, then no span exists for that ID (either it has
+    /// closed or the ID is invalid).
+    #[inline]
+    pub fn metadata(&self, id: &span::Id) -> Option<&'static Metadata<'static>>
+    where
+        S: for<'lookup> LookupSpan<'lookup>,
+    {
+        let span = self.span(id)?;
+        Some(span.metadata())
+    }
+
+    /// Returns [stored data] for the span with the given `id`, if it exists.
+    ///
+    /// If this returns `None`, then no span exists for that ID (either it has
+    /// closed or the ID is invalid).
+    ///
+    /// <pre class="ignore" style="white-space:normal;font:inherit;">
+    ///     <strong>Note</strong>: This requires the wrapped subscriber to
+    ///     implement the <a href="../registry/trait.LookupSpan.html"><code>
+    ///     LookupSpan</code></a> trait. See the documentation on
+    ///     <a href="./struct.Context.html"><code>Context</code>'s
+    ///     declaration</a> for details.
+    /// </pre>
+    ///
+    /// [stored data]: crate::registry::SpanRef
+    #[inline]
+    pub fn span(&self, id: &span::Id) -> Option<registry::SpanRef<'_, S>>
+    where
+        S: for<'lookup> LookupSpan<'lookup>,
+    {
+        let span = self.subscriber.as_ref()?.span(id)?;
+
+        #[cfg(all(feature = "registry", feature = "std"))]
+        return span.try_with_filter(self.filter);
+
+        #[cfg(not(feature = "registry"))]
+        Some(span)
+    }
+
+    /// Returns `true` if an active span exists for the given `Id`.
+    ///
+    /// <pre class="ignore" style="white-space:normal;font:inherit;">
+    ///     <strong>Note</strong>: This requires the wrapped subscriber to
+    ///     implement the <a href="../registry/trait.LookupSpan.html"><code>
+    ///     LookupSpan</code></a> trait. See the documentation on
+    ///     <a href="./struct.Context.html"><code>Context</code>'s
+    ///     declaration</a> for details.
+    /// </pre>
+    #[inline]
+    pub fn exists(&self, id: &span::Id) -> bool
+    where
+        S: for<'lookup> LookupSpan<'lookup>,
+    {
+        self.subscriber.as_ref().and_then(|s| s.span(id)).is_some()
+    }
+
+    /// Returns [stored data] for the span that the wrapped subscriber considers
+    /// to be the current.
+    ///
+    /// If this returns `None`, then we are not currently within a span.
+    ///
+    /// <pre class="ignore" style="white-space:normal;font:inherit;">
+    ///     <strong>Note</strong>: This requires the wrapped subscriber to
+    ///     implement the <a href="../registry/trait.LookupSpan.html"><code>
+    ///     LookupSpan</code></a> trait. See the documentation on
+    ///     <a href="./struct.Context.html"><code>Context</code>'s
+    ///     declaration</a> for details.
+    /// </pre>
+    ///
+    /// [stored data]: crate::registry::SpanRef
+    #[inline]
+    pub fn lookup_current(&self) -> Option<registry::SpanRef<'_, S>>
+    where
+        S: for<'lookup> LookupSpan<'lookup>,
+    {
+        let subscriber = *self.subscriber.as_ref()?;
+        let current = subscriber.current_span();
+        let id = current.id()?;
+        let span = subscriber.span(id);
+        debug_assert!(
+            span.is_some(),
+            "the subscriber should have data for the current span ({:?})!",
+            id,
+        );
+
+        // If we found a span, and our per-layer filter enables it, return that
+        // span!
+        #[cfg(all(feature = "registry", feature = "std"))]
+        {
+            if let Some(span) = span?.try_with_filter(self.filter) {
+                Some(span)
+            } else {
+                // Otherwise, the span at the *top* of the stack is disabled by
+                // per-layer filtering, but there may be additional spans in the stack.
+                //
+                // Currently, `LookupSpan` doesn't have a nice way of exposing access to
+                // the whole span stack. However, if we can downcast the innermost
+                // subscriber to a a `Registry`, we can iterate over its current span
+                // stack.
+                //
+                // TODO(eliza): when https://github.com/tokio-rs/tracing/issues/1459 is
+                // implemented, change this to use that instead...
+                self.lookup_current_filtered(subscriber)
+            }
+        }
+
+        #[cfg(not(feature = "registry"))]
+        span
+    }
+
+    /// Slow path for when the current span is disabled by PLF and we have a
+    /// registry.
+    // This is called by `lookup_current` in the case that per-layer filtering
+    // is in use. `lookup_current` is allowed to be inlined, but this method is
+    // factored out to prevent the loop and (potentially-recursive) subscriber
+    // downcasting from being inlined if `lookup_current` is inlined.
+    #[inline(never)]
+    #[cfg(all(feature = "registry", feature = "std"))]
+    fn lookup_current_filtered<'lookup>(
+        &self,
+        subscriber: &'lookup S,
+    ) -> Option<registry::SpanRef<'lookup, S>>
+    where
+        S: LookupSpan<'lookup>,
+    {
+        let registry = (subscriber as &dyn Subscriber).downcast_ref::<Registry>()?;
+        registry
+            .span_stack()
+            .iter()
+            .find_map(|id| subscriber.span(id)?.try_with_filter(self.filter))
+    }
+
+    /// Returns an iterator over the [stored data] for all the spans in the
+    /// current context, starting with the specified span and ending with the
+    /// root of the trace tree.
+    ///
+    /// <pre class="ignore" style="white-space:normal;font:inherit;">
+    /// <strong>Note</strong>: This returns the spans in reverse order (from leaf to root). Use
+    /// <a href="../registry/struct.Scope.html#method.from_root"><code>Scope::from_root</code></a>
+    /// in case root-to-leaf ordering is desired.
+    /// </pre>
+    ///
+    /// <pre class="ignore" style="white-space:normal;font:inherit;">
+    ///     <strong>Note</strong>: This requires the wrapped subscriber to
+    ///     implement the <a href="../registry/trait.LookupSpan.html"><code>
+    ///     LookupSpan</code></a> trait. See the documentation on
+    ///     <a href="./struct.Context.html"><code>Context</code>'s
+    ///     declaration</a> for details.
+    /// </pre>
+    ///
+    /// [stored data]: crate::registry::SpanRef
+    pub fn span_scope(&self, id: &span::Id) -> Option<registry::Scope<'_, S>>
+    where
+        S: for<'lookup> LookupSpan<'lookup>,
+    {
+        Some(self.span(id)?.scope())
+    }
+
+    /// Returns an iterator over the [stored data] for all the spans in the
+    /// current context, starting with the parent span of the specified event,
+    /// and ending with the root of the trace tree and ending with the current span.
+    ///
+    /// <pre class="ignore" style="white-space:normal;font:inherit;">
+    /// <strong>Note</strong>: Compared to <a href="#method.scope"><code>scope</code></a> this
+    /// returns the spans in reverse order (from leaf to root). Use
+    /// <a href="../registry/struct.Scope.html#method.from_root"><code>Scope::from_root</code></a>
+    /// in case root-to-leaf ordering is desired.
+    /// </pre>
+    ///
+    /// <pre class="ignore" style="white-space:normal;font:inherit;">
+    ///     <strong>Note</strong>: This requires the wrapped subscriber to
+    ///     implement the <a href="../registry/trait.LookupSpan.html"><code>
+    ///     LookupSpan</code></a> trait. See the documentation on
+    ///     <a href="./struct.Context.html"><code>Context</code>'s
+    ///     declaration</a> for details.
+    /// </pre>
+    ///
+    /// [stored data]: crate::registry::SpanRef
+    pub fn event_scope(&self, event: &Event<'_>) -> Option<registry::Scope<'_, S>>
+    where
+        S: for<'lookup> LookupSpan<'lookup>,
+    {
+        Some(self.event_span(event)?.scope())
+    }
+
+    #[cfg(all(feature = "registry", feature = "std"))]
+    pub(crate) fn with_filter(self, filter: FilterId) -> Self {
+        // If we already have our own `FilterId`, combine it with the provided
+        // one. That way, the new `FilterId` will consider a span to be disabled
+        // if it was disabled by the given `FilterId` *or* any `FilterId`s for
+        // layers "above" us in the stack.
+        //
+        // See the doc comment for `FilterId::and` for details.
+        let filter = self.filter.and(filter);
+        Self { filter, ..self }
+    }
+
+    #[cfg(all(feature = "registry", feature = "std"))]
+    pub(crate) fn is_enabled_for(&self, span: &span::Id, filter: FilterId) -> bool
+    where
+        S: for<'lookup> LookupSpan<'lookup>,
+    {
+        self.is_enabled_inner(span, filter).unwrap_or(false)
+    }
+
+    #[cfg(all(feature = "registry", feature = "std"))]
+    pub(crate) fn if_enabled_for(self, span: &span::Id, filter: FilterId) -> Option<Self>
+    where
+        S: for<'lookup> LookupSpan<'lookup>,
+    {
+        if self.is_enabled_inner(span, filter)? {
+            Some(self.with_filter(filter))
+        } else {
+            None
+        }
+    }
+
+    #[cfg(all(feature = "registry", feature = "std"))]
+    fn is_enabled_inner(&self, span: &span::Id, filter: FilterId) -> Option<bool>
+    where
+        S: for<'lookup> LookupSpan<'lookup>,
+    {
+        Some(self.span(span)?.is_enabled_for(filter))
+    }
+}
+
+impl<S> Context<'_, S> {
+    pub(crate) fn none() -> Self {
+        Self {
+            subscriber: None,
+
+            #[cfg(feature = "registry")]
+            filter: FilterId::none(),
+        }
+    }
+}
+
+impl<S> Clone for Context<'_, S> {
+    #[inline]
+    fn clone(&self) -> Self {
+        let subscriber = self.subscriber.as_ref().copied();
+        Context {
+            subscriber,
+
+            #[cfg(all(feature = "registry", feature = "std"))]
+            filter: self.filter,
+        }
+    }
+}

vendor/tracing-subscriber/src/layer/layered.rs

@@ -0,0 +1,554 @@
+use tracing_core::{metadata::Metadata, span, Dispatch, Event, Interest, LevelFilter, Subscriber};
+
+use crate::{
+    filter,
+    layer::{Context, Layer},
+    registry::LookupSpan,
+};
+#[cfg(all(feature = "registry", feature = "std"))]
+use crate::{filter::FilterId, registry::Registry};
+use core::{
+    any::{Any, TypeId},
+    cmp, fmt,
+    marker::PhantomData,
+};
+
+/// A [`Subscriber`] composed of a `Subscriber` wrapped by one or more
+/// [`Layer`]s.
+///
+/// [`Layer`]: crate::Layer
+/// [`Subscriber`]: tracing_core::Subscriber
+#[derive(Clone)]
+pub struct Layered<L, I, S = I> {
+    /// The layer.
+    layer: L,
+
+    /// The inner value that `self.layer` was layered onto.
+    ///
+    /// If this is also a `Layer`, then this `Layered` will implement `Layer`.
+    /// If this is a `Subscriber`, then this `Layered` will implement
+    /// `Subscriber` instead.
+    inner: I,
+
+    // These booleans are used to determine how to combine `Interest`s and max
+    // level hints when per-layer filters are in use.
+    /// Is `self.inner` a `Registry`?
+    ///
+    /// If so, when combining `Interest`s, we want to "bubble up" its
+    /// `Interest`.
+    inner_is_registry: bool,
+
+    /// Does `self.layer` have per-layer filters?
+    ///
+    /// This will be true if:
+    /// - `self.inner` is a `Filtered`.
+    /// - `self.inner` is a tree of `Layered`s where _all_ arms of those
+    ///   `Layered`s have per-layer filters.
+    ///
+    /// Otherwise, if it's a `Layered` with one per-layer filter in one branch,
+    /// but a non-per-layer-filtered layer in the other branch, this will be
+    /// _false_, because the `Layered` is already handling the combining of
+    /// per-layer filter `Interest`s and max level hints with its non-filtered
+    /// `Layer`.
+    has_layer_filter: bool,
+
+    /// Does `self.inner` have per-layer filters?
+    ///
+    /// This is determined according to the same rules as
+    /// `has_layer_filter` above.
+    inner_has_layer_filter: bool,
+    _s: PhantomData<fn(S)>,
+}
+
+// === impl Layered ===
+
+impl<L, S> Layered<L, S>
+where
+    L: Layer<S>,
+    S: Subscriber,
+{
+    /// Returns `true` if this [`Subscriber`] is the same type as `T`.
+    pub fn is<T: Any>(&self) -> bool {
+        self.downcast_ref::<T>().is_some()
+    }
+
+    /// Returns some reference to this [`Subscriber`] value if it is of type `T`,
+    /// or `None` if it isn't.
+    pub fn downcast_ref<T: Any>(&self) -> Option<&T> {
+        unsafe {
+            let raw = self.downcast_raw(TypeId::of::<T>())?;
+            if raw.is_null() {
+                None
+            } else {
+                Some(&*(raw as *const T))
+            }
+        }
+    }
+}
+
+impl<L, S> Subscriber for Layered<L, S>
+where
+    L: Layer<S>,
+    S: Subscriber,
+{
+    fn register_callsite(&self, metadata: &'static Metadata<'static>) -> Interest {
+        self.pick_interest(self.layer.register_callsite(metadata), || {
+            self.inner.register_callsite(metadata)
+        })
+    }
+
+    fn enabled(&self, metadata: &Metadata<'_>) -> bool {
+        if self.layer.enabled(metadata, self.ctx()) {
+            // if the outer layer enables the callsite metadata, ask the subscriber.
+            self.inner.enabled(metadata)
+        } else {
+            // otherwise, the callsite is disabled by the layer
+
+            // If per-layer filters are in use, and we are short-circuiting
+            // (rather than calling into the inner type), clear the current
+            // per-layer filter `enabled` state.
+            #[cfg(feature = "registry")]
+            filter::FilterState::clear_enabled();
+
+            false
+        }
+    }
+
+    fn max_level_hint(&self) -> Option<LevelFilter> {
+        self.pick_level_hint(
+            self.layer.max_level_hint(),
+            self.inner.max_level_hint(),
+            super::subscriber_is_none(&self.inner),
+        )
+    }
+
+    fn new_span(&self, span: &span::Attributes<'_>) -> span::Id {
+        let id = self.inner.new_span(span);
+        self.layer.on_new_span(span, &id, self.ctx());
+        id
+    }
+
+    fn record(&self, span: &span::Id, values: &span::Record<'_>) {
+        self.inner.record(span, values);
+        self.layer.on_record(span, values, self.ctx());
+    }
+
+    fn record_follows_from(&self, span: &span::Id, follows: &span::Id) {
+        self.inner.record_follows_from(span, follows);
+        self.layer.on_follows_from(span, follows, self.ctx());
+    }
+
+    fn event_enabled(&self, event: &Event<'_>) -> bool {
+        if self.layer.event_enabled(event, self.ctx()) {
+            // if the outer layer enables the event, ask the inner subscriber.
+            self.inner.event_enabled(event)
+        } else {
+            // otherwise, the event is disabled by this layer
+            false
+        }
+    }
+
+    fn event(&self, event: &Event<'_>) {
+        self.inner.event(event);
+        self.layer.on_event(event, self.ctx());
+    }
+
+    fn enter(&self, span: &span::Id) {
+        self.inner.enter(span);
+        self.layer.on_enter(span, self.ctx());
+    }
+
+    fn exit(&self, span: &span::Id) {
+        self.inner.exit(span);
+        self.layer.on_exit(span, self.ctx());
+    }
+
+    fn clone_span(&self, old: &span::Id) -> span::Id {
+        let new = self.inner.clone_span(old);
+        if &new != old {
+            self.layer.on_id_change(old, &new, self.ctx())
+        };
+        new
+    }
+
+    #[inline]
+    fn drop_span(&self, id: span::Id) {
+        self.try_close(id);
+    }
+
+    fn try_close(&self, id: span::Id) -> bool {
+        #[cfg(all(feature = "registry", feature = "std"))]
+        let subscriber = &self.inner as &dyn Subscriber;
+        #[cfg(all(feature = "registry", feature = "std"))]
+        let mut guard = subscriber
+            .downcast_ref::<Registry>()
+            .map(|registry| registry.start_close(id.clone()));
+        if self.inner.try_close(id.clone()) {
+            // If we have a registry's close guard, indicate that the span is
+            // closing.
+            #[cfg(all(feature = "registry", feature = "std"))]
+            {
+                if let Some(g) = guard.as_mut() {
+                    g.set_closing()
+                };
+            }
+
+            self.layer.on_close(id, self.ctx());
+            true
+        } else {
+            false
+        }
+    }
+
+    #[inline]
+    fn current_span(&self) -> span::Current {
+        self.inner.current_span()
+    }
+
+    #[doc(hidden)]
+    unsafe fn downcast_raw(&self, id: TypeId) -> Option<*const ()> {
+        // Unlike the implementation of `Layer` for `Layered`, we don't have to
+        // handle the "magic PLF downcast marker" here. If a `Layered`
+        // implements `Subscriber`, we already know that the `inner` branch is
+        // going to contain something that doesn't have per-layer filters (the
+        // actual root `Subscriber`). Thus, a `Layered` that implements
+        // `Subscriber` will always be propagating the root subscriber's
+        // `Interest`/level hint, even if it includes a `Layer` that has
+        // per-layer filters, because it will only ever contain layers where
+        // _one_ child has per-layer filters.
+        //
+        // The complex per-layer filter detection logic is only relevant to
+        // *trees* of layers, which involve the `Layer` implementation for
+        // `Layered`, not *lists* of layers, where every `Layered` implements
+        // `Subscriber`. Of course, a linked list can be thought of as a
+        // degenerate tree...but luckily, we are able to make a type-level
+        // distinction between individual `Layered`s that are definitely
+        // list-shaped (their inner child implements `Subscriber`), and
+        // `Layered`s that might be tree-shaped (the inner child is also a
+        // `Layer`).
+
+        // If downcasting to `Self`, return a pointer to `self`.
+        if id == TypeId::of::<Self>() {
+            return Some(self as *const _ as *const ());
+        }
+
+        self.layer
+            .downcast_raw(id)
+            .or_else(|| self.inner.downcast_raw(id))
+    }
+}
+
+impl<S, A, B> Layer<S> for Layered<A, B, S>
+where
+    A: Layer<S>,
+    B: Layer<S>,
+    S: Subscriber,
+{
+    fn on_register_dispatch(&self, subscriber: &Dispatch) {
+        self.layer.on_register_dispatch(subscriber);
+        self.inner.on_register_dispatch(subscriber);
+    }
+
+    fn on_layer(&mut self, subscriber: &mut S) {
+        self.layer.on_layer(subscriber);
+        self.inner.on_layer(subscriber);
+    }
+
+    fn register_callsite(&self, metadata: &'static Metadata<'static>) -> Interest {
+        self.pick_interest(self.layer.register_callsite(metadata), || {
+            self.inner.register_callsite(metadata)
+        })
+    }
+
+    fn enabled(&self, metadata: &Metadata<'_>, ctx: Context<'_, S>) -> bool {
+        if self.layer.enabled(metadata, ctx.clone()) {
+            // if the outer subscriber enables the callsite metadata, ask the inner layer.
+            self.inner.enabled(metadata, ctx)
+        } else {
+            // otherwise, the callsite is disabled by this layer
+            false
+        }
+    }
+
+    fn max_level_hint(&self) -> Option<LevelFilter> {
+        self.pick_level_hint(
+            self.layer.max_level_hint(),
+            self.inner.max_level_hint(),
+            super::layer_is_none(&self.inner),
+        )
+    }
+
+    #[inline]
+    fn on_new_span(&self, attrs: &span::Attributes<'_>, id: &span::Id, ctx: Context<'_, S>) {
+        self.inner.on_new_span(attrs, id, ctx.clone());
+        self.layer.on_new_span(attrs, id, ctx);
+    }
+
+    #[inline]
+    fn on_record(&self, span: &span::Id, values: &span::Record<'_>, ctx: Context<'_, S>) {
+        self.inner.on_record(span, values, ctx.clone());
+        self.layer.on_record(span, values, ctx);
+    }
+
+    #[inline]
+    fn on_follows_from(&self, span: &span::Id, follows: &span::Id, ctx: Context<'_, S>) {
+        self.inner.on_follows_from(span, follows, ctx.clone());
+        self.layer.on_follows_from(span, follows, ctx);
+    }
+
+    #[inline]
+    fn event_enabled(&self, event: &Event<'_>, ctx: Context<'_, S>) -> bool {
+        if self.layer.event_enabled(event, ctx.clone()) {
+            // if the outer layer enables the event, ask the inner subscriber.
+            self.inner.event_enabled(event, ctx)
+        } else {
+            // otherwise, the event is disabled by this layer
+            false
+        }
+    }
+
+    #[inline]
+    fn on_event(&self, event: &Event<'_>, ctx: Context<'_, S>) {
+        self.inner.on_event(event, ctx.clone());
+        self.layer.on_event(event, ctx);
+    }
+
+    #[inline]
+    fn on_enter(&self, id: &span::Id, ctx: Context<'_, S>) {
+        self.inner.on_enter(id, ctx.clone());
+        self.layer.on_enter(id, ctx);
+    }
+
+    #[inline]
+    fn on_exit(&self, id: &span::Id, ctx: Context<'_, S>) {
+        self.inner.on_exit(id, ctx.clone());
+        self.layer.on_exit(id, ctx);
+    }
+
+    #[inline]
+    fn on_close(&self, id: span::Id, ctx: Context<'_, S>) {
+        self.inner.on_close(id.clone(), ctx.clone());
+        self.layer.on_close(id, ctx);
+    }
+
+    #[inline]
+    fn on_id_change(&self, old: &span::Id, new: &span::Id, ctx: Context<'_, S>) {
+        self.inner.on_id_change(old, new, ctx.clone());
+        self.layer.on_id_change(old, new, ctx);
+    }
+
+    #[doc(hidden)]
+    unsafe fn downcast_raw(&self, id: TypeId) -> Option<*const ()> {
+        match id {
+            // If downcasting to `Self`, return a pointer to `self`.
+            id if id == TypeId::of::<Self>() => Some(self as *const _ as *const ()),
+
+            // Oh, we're looking for per-layer filters!
+            //
+            // This should only happen if we are inside of another `Layered`,
+            // and it's trying to determine how it should combine `Interest`s
+            // and max level hints.
+            //
+            // In that case, this `Layered` should be considered to be
+            // "per-layer filtered" if *both* the outer layer and the inner
+            // layer/subscriber have per-layer filters. Otherwise, this `Layered
+            // should *not* be considered per-layer filtered (even if one or the
+            // other has per layer filters). If only one `Layer` is per-layer
+            // filtered, *this* `Layered` will handle aggregating the `Interest`
+            // and level hints on behalf of its children, returning the
+            // aggregate (which is the value from the &non-per-layer-filtered*
+            // child).
+            //
+            // Yes, this rule *is* slightly counter-intuitive, but it's
+            // necessary due to a weird edge case that can occur when two
+            // `Layered`s where one side is per-layer filtered and the other
+            // isn't are `Layered` together to form a tree. If we didn't have
+            // this rule, we would actually end up *ignoring* `Interest`s from
+            // the non-per-layer-filtered layers, since both branches would
+            // claim to have PLF.
+            //
+            // If you don't understand this...that's fine, just don't mess with
+            // it. :)
+            id if filter::is_plf_downcast_marker(id) => {
+                self.layer.downcast_raw(id).and(self.inner.downcast_raw(id))
+            }
+
+            // Otherwise, try to downcast both branches normally...
+            _ => self
+                .layer
+                .downcast_raw(id)
+                .or_else(|| self.inner.downcast_raw(id)),
+        }
+    }
+}
+
+impl<'a, L, S> LookupSpan<'a> for Layered<L, S>
+where
+    S: Subscriber + LookupSpan<'a>,
+{
+    type Data = S::Data;
+
+    fn span_data(&'a self, id: &span::Id) -> Option<Self::Data> {
+        self.inner.span_data(id)
+    }
+
+    #[cfg(all(feature = "registry", feature = "std"))]
+    fn register_filter(&mut self) -> FilterId {
+        self.inner.register_filter()
+    }
+}
+
+impl<L, S> Layered<L, S>
+where
+    S: Subscriber,
+{
+    fn ctx(&self) -> Context<'_, S> {
+        Context::new(&self.inner)
+    }
+}
+
+impl<A, B, S> Layered<A, B, S>
+where
+    A: Layer<S>,
+    S: Subscriber,
+{
+    pub(super) fn new(layer: A, inner: B, inner_has_layer_filter: bool) -> Self {
+        #[cfg(all(feature = "registry", feature = "std"))]
+        let inner_is_registry = TypeId::of::<S>() == TypeId::of::<crate::registry::Registry>();
+
+        #[cfg(not(all(feature = "registry", feature = "std")))]
+        let inner_is_registry = false;
+
+        let inner_has_layer_filter = inner_has_layer_filter || inner_is_registry;
+        let has_layer_filter = filter::layer_has_plf(&layer);
+        Self {
+            layer,
+            inner,
+            has_layer_filter,
+            inner_has_layer_filter,
+            inner_is_registry,
+            _s: PhantomData,
+        }
+    }
+
+    fn pick_interest(&self, outer: Interest, inner: impl FnOnce() -> Interest) -> Interest {
+        if self.has_layer_filter {
+            return inner();
+        }
+
+        // If the outer layer has disabled the callsite, return now so that
+        // the inner layer/subscriber doesn't get its hopes up.
+        if outer.is_never() {
+            // If per-layer filters are in use, and we are short-circuiting
+            // (rather than calling into the inner type), clear the current
+            // per-layer filter interest state.
+            #[cfg(feature = "registry")]
+            filter::FilterState::take_interest();
+
+            return outer;
+        }
+
+        // The `inner` closure will call `inner.register_callsite()`. We do this
+        // before the `if` statement to  ensure that the inner subscriber is
+        // informed that the callsite exists regardless of the outer layer's
+        // filtering decision.
+        let inner = inner();
+        if outer.is_sometimes() {
+            // if this interest is "sometimes", return "sometimes" to ensure that
+            // filters are reevaluated.
+            return outer;
+        }
+
+        // If there is a per-layer filter in the `inner` stack, and it returns
+        // `never`, change the interest to `sometimes`, because the `outer`
+        // layer didn't return `never`. This means that _some_ layer still wants
+        // to see that callsite, even though the inner stack's per-layer filter
+        // didn't want it. Therefore, returning `sometimes` will ensure
+        // `enabled` is called so that the per-layer filter can skip that
+        // span/event, while the `outer` layer still gets to see it.
+        if inner.is_never() && self.inner_has_layer_filter {
+            return Interest::sometimes();
+        }
+
+        // otherwise, allow the inner subscriber or subscriber to weigh in.
+        inner
+    }
+
+    fn pick_level_hint(
+        &self,
+        outer_hint: Option<LevelFilter>,
+        inner_hint: Option<LevelFilter>,
+        inner_is_none: bool,
+    ) -> Option<LevelFilter> {
+        if self.inner_is_registry {
+            return outer_hint;
+        }
+
+        if self.has_layer_filter && self.inner_has_layer_filter {
+            return Some(cmp::max(outer_hint?, inner_hint?));
+        }
+
+        if self.has_layer_filter && inner_hint.is_none() {
+            return None;
+        }
+
+        if self.inner_has_layer_filter && outer_hint.is_none() {
+            return None;
+        }
+
+        // If the layer is `Option::None`, then we
+        // want to short-circuit the layer underneath, if it
+        // returns `None`, to override the `None` layer returning
+        // `Some(OFF)`, which should ONLY apply when there are
+        // no other layers that return `None`. Note this
+        // `None` does not == `Some(TRACE)`, it means
+        // something more like: "whatever all the other
+        // layers agree on, default to `TRACE` if none
+        // have an opinion". We also choose do this AFTER
+        // we check for per-layer filters, which
+        // have their own logic.
+        //
+        // Also note that this does come at some perf cost, but
+        // this function is only called on initialization and
+        // subscriber reloading.
+        if super::layer_is_none(&self.layer) {
+            return cmp::max(outer_hint, Some(inner_hint?));
+        }
+
+        // Similarly, if the layer on the inside is `None` and it returned an
+        // `Off` hint, we want to override that with the outer hint.
+        if inner_is_none && inner_hint == Some(LevelFilter::OFF) {
+            return outer_hint;
+        }
+
+        cmp::max(outer_hint, inner_hint)
+    }
+}
+
+impl<A, B, S> fmt::Debug for Layered<A, B, S>
+where
+    A: fmt::Debug,
+    B: fmt::Debug,
+{
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        #[cfg(all(feature = "registry", feature = "std"))]
+        let alt = f.alternate();
+        let mut s = f.debug_struct("Layered");
+        // These additional fields are more verbose and usually only necessary
+        // for internal debugging purposes, so only print them if alternate mode
+        // is enabled.
+
+        #[cfg(all(feature = "registry", feature = "std"))]
+        {
+            if alt {
+                s.field("inner_is_registry", &self.inner_is_registry)
+                    .field("has_layer_filter", &self.has_layer_filter)
+                    .field("inner_has_layer_filter", &self.inner_has_layer_filter);
+            }
+        }
+
+        s.field("layer", &self.layer)
+            .field("inner", &self.inner)
+            .finish()
+    }
+}

vendor/tracing-subscriber/src/layer/tests.rs

@@ -0,0 +1,308 @@
+use super::*;
+use tracing_core::subscriber::NoSubscriber;
+
+#[derive(Debug)]
+pub(crate) struct NopLayer;
+impl<S: Subscriber> Layer<S> for NopLayer {}
+
+#[allow(dead_code)]
+struct NopLayer2;
+impl<S: Subscriber> Layer<S> for NopLayer2 {}
+
+/// A layer that holds a string.
+///
+/// Used to test that pointers returned by downcasting are actually valid.
+struct StringLayer(&'static str);
+impl<S: Subscriber> Layer<S> for StringLayer {}
+struct StringLayer2(&'static str);
+impl<S: Subscriber> Layer<S> for StringLayer2 {}
+
+struct StringLayer3(&'static str);
+impl<S: Subscriber> Layer<S> for StringLayer3 {}
+
+pub(crate) struct StringSubscriber(&'static str);
+
+impl Subscriber for StringSubscriber {
+    fn register_callsite(&self, _: &'static Metadata<'static>) -> Interest {
+        Interest::never()
+    }
+
+    fn enabled(&self, _: &Metadata<'_>) -> bool {
+        false
+    }
+
+    fn new_span(&self, _: &span::Attributes<'_>) -> span::Id {
+        span::Id::from_u64(1)
+    }
+
+    fn record(&self, _: &span::Id, _: &span::Record<'_>) {}
+    fn record_follows_from(&self, _: &span::Id, _: &span::Id) {}
+    fn event(&self, _: &Event<'_>) {}
+    fn enter(&self, _: &span::Id) {}
+    fn exit(&self, _: &span::Id) {}
+}
+
+fn assert_subscriber(_s: impl Subscriber) {}
+fn assert_layer<S: Subscriber>(_l: &impl Layer<S>) {}
+
+#[test]
+fn layer_is_subscriber() {
+    let s = NopLayer.with_subscriber(NoSubscriber::default());
+    assert_subscriber(s)
+}
+
+#[test]
+fn two_layers_are_subscriber() {
+    let s = NopLayer
+        .and_then(NopLayer)
+        .with_subscriber(NoSubscriber::default());
+    assert_subscriber(s)
+}
+
+#[test]
+fn three_layers_are_subscriber() {
+    let s = NopLayer
+        .and_then(NopLayer)
+        .and_then(NopLayer)
+        .with_subscriber(NoSubscriber::default());
+    assert_subscriber(s)
+}
+
+#[test]
+fn three_layers_are_layer() {
+    let layers = NopLayer.and_then(NopLayer).and_then(NopLayer);
+    assert_layer(&layers);
+    let _ = layers.with_subscriber(NoSubscriber::default());
+}
+
+#[test]
+#[cfg(feature = "alloc")]
+fn box_layer_is_layer() {
+    use alloc::boxed::Box;
+    let l: Box<dyn Layer<NoSubscriber> + Send + Sync> = Box::new(NopLayer);
+    assert_layer(&l);
+    l.with_subscriber(NoSubscriber::default());
+}
+
+#[test]
+fn downcasts_to_subscriber() {
+    let s = NopLayer
+        .and_then(NopLayer)
+        .and_then(NopLayer)
+        .with_subscriber(StringSubscriber("subscriber"));
+    let subscriber =
+        <dyn Subscriber>::downcast_ref::<StringSubscriber>(&s).expect("subscriber should downcast");
+    assert_eq!(subscriber.0, "subscriber");
+}
+
+#[test]
+fn downcasts_to_layer() {
+    let s = StringLayer("layer_1")
+        .and_then(StringLayer2("layer_2"))
+        .and_then(StringLayer3("layer_3"))
+        .with_subscriber(NoSubscriber::default());
+    let layer = <dyn Subscriber>::downcast_ref::<StringLayer>(&s).expect("layer 1 should downcast");
+    assert_eq!(layer.0, "layer_1");
+    let layer =
+        <dyn Subscriber>::downcast_ref::<StringLayer2>(&s).expect("layer 2 should downcast");
+    assert_eq!(layer.0, "layer_2");
+    let layer =
+        <dyn Subscriber>::downcast_ref::<StringLayer3>(&s).expect("layer 3 should downcast");
+    assert_eq!(layer.0, "layer_3");
+}
+
+#[cfg(all(feature = "registry", feature = "std"))]
+mod registry_tests {
+    use super::*;
+    use crate::registry::LookupSpan;
+
+    #[test]
+    fn context_event_span() {
+        use std::sync::{Arc, Mutex};
+        let last_event_span = Arc::new(Mutex::new(None));
+
+        struct RecordingLayer {
+            last_event_span: Arc<Mutex<Option<&'static str>>>,
+        }
+
+        impl<S> Layer<S> for RecordingLayer
+        where
+            S: Subscriber + for<'lookup> LookupSpan<'lookup>,
+        {
+            fn on_event(&self, event: &Event<'_>, ctx: Context<'_, S>) {
+                let span = ctx.event_span(event);
+                *self.last_event_span.lock().unwrap() = span.map(|s| s.name());
+            }
+        }
+
+        tracing::subscriber::with_default(
+            crate::registry().with(RecordingLayer {
+                last_event_span: last_event_span.clone(),
+            }),
+            || {
+                tracing::info!("no span");
+                assert_eq!(*last_event_span.lock().unwrap(), None);
+
+                let parent = tracing::info_span!("explicit");
+                tracing::info!(parent: &parent, "explicit span");
+                assert_eq!(*last_event_span.lock().unwrap(), Some("explicit"));
+
+                let _guard = tracing::info_span!("contextual").entered();
+                tracing::info!("contextual span");
+                assert_eq!(*last_event_span.lock().unwrap(), Some("contextual"));
+            },
+        );
+    }
+
+    /// Tests for how max-level hints are calculated when combining layers
+    /// with and without per-layer filtering.
+    mod max_level_hints {
+
+        use super::*;
+        use crate::filter::*;
+
+        #[test]
+        fn mixed_with_unfiltered() {
+            let subscriber = crate::registry()
+                .with(NopLayer)
+                .with(NopLayer.with_filter(LevelFilter::INFO));
+            assert_eq!(subscriber.max_level_hint(), None);
+        }
+
+        #[test]
+        fn mixed_with_unfiltered_layered() {
+            let subscriber = crate::registry().with(NopLayer).with(
+                NopLayer
+                    .with_filter(LevelFilter::INFO)
+                    .and_then(NopLayer.with_filter(LevelFilter::TRACE)),
+            );
+            assert_eq!(dbg!(subscriber).max_level_hint(), None);
+        }
+
+        #[test]
+        fn mixed_interleaved() {
+            let subscriber = crate::registry()
+                .with(NopLayer)
+                .with(NopLayer.with_filter(LevelFilter::INFO))
+                .with(NopLayer)
+                .with(NopLayer.with_filter(LevelFilter::INFO));
+            assert_eq!(dbg!(subscriber).max_level_hint(), None);
+        }
+
+        #[test]
+        fn mixed_layered() {
+            let subscriber = crate::registry()
+                .with(NopLayer.with_filter(LevelFilter::INFO).and_then(NopLayer))
+                .with(NopLayer.and_then(NopLayer.with_filter(LevelFilter::INFO)));
+            assert_eq!(dbg!(subscriber).max_level_hint(), None);
+        }
+
+        #[test]
+        fn plf_only_unhinted() {
+            let subscriber = crate::registry()
+                .with(NopLayer.with_filter(LevelFilter::INFO))
+                .with(NopLayer.with_filter(filter_fn(|_| true)));
+            assert_eq!(dbg!(subscriber).max_level_hint(), None);
+        }
+
+        #[test]
+        fn plf_only_unhinted_nested_outer() {
+            // if a nested tree of per-layer filters has an _outer_ filter with
+            // no max level hint, it should return `None`.
+            let subscriber = crate::registry()
+                .with(
+                    NopLayer
+                        .with_filter(LevelFilter::INFO)
+                        .and_then(NopLayer.with_filter(LevelFilter::WARN)),
+                )
+                .with(
+                    NopLayer
+                        .with_filter(filter_fn(|_| true))
+                        .and_then(NopLayer.with_filter(LevelFilter::DEBUG)),
+                );
+            assert_eq!(dbg!(subscriber).max_level_hint(), None);
+        }
+
+        #[test]
+        fn plf_only_unhinted_nested_inner() {
+            // If a nested tree of per-layer filters has an _inner_ filter with
+            // no max-level hint, but the _outer_ filter has a max level hint,
+            // it should pick the outer hint. This is because the outer filter
+            // will disable the spans/events before they make it to the inner
+            // filter.
+            let subscriber = dbg!(crate::registry().with(
+                NopLayer
+                    .with_filter(filter_fn(|_| true))
+                    .and_then(NopLayer.with_filter(filter_fn(|_| true)))
+                    .with_filter(LevelFilter::INFO),
+            ));
+            assert_eq!(dbg!(subscriber).max_level_hint(), Some(LevelFilter::INFO));
+        }
+
+        #[test]
+        fn unhinted_nested_inner() {
+            let subscriber = dbg!(crate::registry()
+                .with(NopLayer.and_then(NopLayer).with_filter(LevelFilter::INFO))
+                .with(
+                    NopLayer
+                        .with_filter(filter_fn(|_| true))
+                        .and_then(NopLayer.with_filter(filter_fn(|_| true)))
+                        .with_filter(LevelFilter::WARN),
+                ));
+            assert_eq!(dbg!(subscriber).max_level_hint(), Some(LevelFilter::INFO));
+        }
+
+        #[test]
+        fn unhinted_nested_inner_mixed() {
+            let subscriber = dbg!(crate::registry()
+                .with(
+                    NopLayer
+                        .and_then(NopLayer.with_filter(filter_fn(|_| true)))
+                        .with_filter(LevelFilter::INFO)
+                )
+                .with(
+                    NopLayer
+                        .with_filter(filter_fn(|_| true))
+                        .and_then(NopLayer.with_filter(filter_fn(|_| true)))
+                        .with_filter(LevelFilter::WARN),
+                ));
+            assert_eq!(dbg!(subscriber).max_level_hint(), Some(LevelFilter::INFO));
+        }
+
+        #[test]
+        fn plf_only_picks_max() {
+            let subscriber = crate::registry()
+                .with(NopLayer.with_filter(LevelFilter::WARN))
+                .with(NopLayer.with_filter(LevelFilter::DEBUG));
+            assert_eq!(dbg!(subscriber).max_level_hint(), Some(LevelFilter::DEBUG));
+        }
+
+        #[test]
+        fn many_plf_only_picks_max() {
+            let subscriber = crate::registry()
+                .with(NopLayer.with_filter(LevelFilter::WARN))
+                .with(NopLayer.with_filter(LevelFilter::DEBUG))
+                .with(NopLayer.with_filter(LevelFilter::INFO))
+                .with(NopLayer.with_filter(LevelFilter::ERROR));
+            assert_eq!(dbg!(subscriber).max_level_hint(), Some(LevelFilter::DEBUG));
+        }
+
+        #[test]
+        fn nested_plf_only_picks_max() {
+            let subscriber = crate::registry()
+                .with(
+                    NopLayer.with_filter(LevelFilter::INFO).and_then(
+                        NopLayer
+                            .with_filter(LevelFilter::WARN)
+                            .and_then(NopLayer.with_filter(LevelFilter::DEBUG)),
+                    ),
+                )
+                .with(
+                    NopLayer
+                        .with_filter(LevelFilter::INFO)
+                        .and_then(NopLayer.with_filter(LevelFilter::ERROR)),
+                );
+            assert_eq!(dbg!(subscriber).max_level_hint(), Some(LevelFilter::DEBUG));
+        }
+    }
+}

vendor/tracing-subscriber/src/lib.rs

@@ -0,0 +1,252 @@
+//! Utilities for implementing and composing [`tracing`] subscribers.
+//!
+//! [`tracing`] is a framework for instrumenting Rust programs to collect
+//! scoped, structured, and async-aware diagnostics. The [`Subscriber`] trait
+//! represents the functionality necessary to collect this trace data. This
+//! crate contains tools for composing subscribers out of smaller units of
+//! behaviour, and batteries-included implementations of common subscriber
+//! functionality.
+//!
+//! `tracing-subscriber` is intended for use by both `Subscriber` authors and
+//! application authors using `tracing` to instrument their applications.
+//!
+//! *Compiler support: [requires `rustc` 1.65+][msrv]*
+//!
+//! [msrv]: #supported-rust-versions
+//!
+//! ## `Layer`s and `Filter`s
+//!
+//! The most important component of the `tracing-subscriber` API is the
+//! [`Layer`] trait, which provides a composable abstraction for building
+//! [`Subscriber`]s. Like the [`Subscriber`] trait, a [`Layer`] defines a
+//! particular behavior for collecting trace data. Unlike [`Subscriber`]s,
+//! which implement a *complete* strategy for how trace data is collected,
+//! [`Layer`]s provide *modular* implementations of specific behaviors.
+//! Therefore, they can be [composed together] to form a [`Subscriber`] which is
+//! capable of recording traces in a variety of ways. See the [`layer` module's
+//! documentation][layer] for details on using [`Layer`]s.
+//!
+//! In addition, the [`Filter`] trait defines an interface for filtering what
+//! spans and events are recorded by a particular layer. This allows different
+//! [`Layer`]s to handle separate subsets of the trace data emitted by a
+//! program. See the [documentation on per-layer filtering][plf] for more
+//! information on using [`Filter`]s.
+//!
+//! [`Layer`]: crate::layer::Layer
+//! [composed together]: crate::layer#composing-layers
+//! [layer]: crate::layer
+//! [`Filter`]: crate::layer::Filter
+//! [plf]: crate::layer#per-layer-filtering
+//!
+//! ## Included Subscribers
+//!
+//! The following `Subscriber`s are provided for application authors:
+//!
+//! - [`fmt`] - Formats and logs tracing data (requires the `fmt` feature flag)
+//!
+//! ## Feature Flags
+//!
+//! - `std`: Enables APIs that depend on the Rust standard library
+//!   (enabled by default).
+//! - `alloc`: Depend on [`liballoc`] (enabled by "std").
+//! - `env-filter`: Enables the [`EnvFilter`] type, which implements filtering
+//!   similar to the [`env_logger` crate]. **Requires "std"**.
+//! - `fmt`: Enables the [`fmt`] module, which provides a subscriber
+//!   implementation for printing formatted representations of trace events.
+//!   Enabled by default. **Requires "registry" and "std"**.
+//! - `ansi`: Enables `fmt` support for ANSI terminal colors. Enabled by
+//!   default.
+//! - `registry`: enables the [`registry`] module. Enabled by default.
+//!   **Requires "std"**.
+//! - `json`: Enables `fmt` support for JSON output. In JSON output, the ANSI
+//!   feature does nothing. **Requires "fmt" and "std"**.
+//! - `local-time`: Enables local time formatting when using the [`time`
+//!   crate]'s timestamp formatters with the `fmt` subscriber.
+//!
+//! [`registry`]: mod@registry
+//!
+//! ### Optional Dependencies
+//!
+//! - [`tracing-log`]: Enables better formatting for events emitted by `log`
+//!   macros in the `fmt` subscriber. Enabled by default.
+//! - [`time`][`time` crate]: Enables support for using the [`time` crate] for timestamp
+//!   formatting in the `fmt` subscriber.
+//! - [`smallvec`]: Causes the `EnvFilter` type to use the `smallvec` crate (rather
+//!   than `Vec`) as a performance optimization. Enabled by default.
+//! - [`parking_lot`]: Use the `parking_lot` crate's `RwLock` implementation
+//!   rather than the Rust standard library's implementation.
+//!
+//! ### `no_std` Support
+//!
+//! In embedded systems and other bare-metal applications, `tracing` can be
+//! used without requiring the Rust standard library, although some features are
+//! disabled. Although most of the APIs provided by `tracing-subscriber`, such
+//! as [`fmt`] and [`EnvFilter`], require the standard library, some
+//! functionality, such as the [`Layer`] trait, can still be used in
+//! `no_std` environments.
+//!
+//! The dependency on the standard library is controlled by two crate feature
+//! flags, "std", which enables the dependency on [`libstd`], and "alloc", which
+//! enables the dependency on [`liballoc`] (and is enabled by the "std"
+//! feature). These features are enabled by default, but `no_std` users can
+//! disable them using:
+//!
+//! ```toml
+//! # Cargo.toml
+//! tracing-subscriber = { version = "0.3", default-features = false }
+//! ```
+//!
+//! Additional APIs are available when [`liballoc`] is available. To enable
+//! `liballoc` but not `std`, use:
+//!
+//! ```toml
+//! # Cargo.toml
+//! tracing-subscriber = { version = "0.3", default-features = false, features = ["alloc"] }
+//! ```
+//!
+//! ### Unstable Features
+//!
+//! These feature flags enable **unstable** features. The public API may break in 0.1.x
+//! releases. To enable these features, the `--cfg tracing_unstable` must be passed to
+//! `rustc` when compiling.
+//!
+//! The following unstable feature flags are currently available:
+//!
+//! * `valuable`: Enables support for serializing values recorded using the
+//!   [`valuable`] crate as structured JSON in the [`format::Json`] formatter.
+//!
+//! #### Enabling Unstable Features
+//!
+//! The easiest way to set the `tracing_unstable` cfg is to use the `RUSTFLAGS`
+//! env variable when running `cargo` commands:
+//!
+//! ```shell
+//! RUSTFLAGS="--cfg tracing_unstable" cargo build
+//! ```
+//! Alternatively, the following can be added to the `.cargo/config` file in a
+//! project to automatically enable the cfg flag for that project:
+//!
+//! ```toml
+//! [build]
+//! rustflags = ["--cfg", "tracing_unstable"]
+//! ```
+//!
+//! [feature flags]: https://doc.rust-lang.org/cargo/reference/manifest.html#the-features-section
+//! [`valuable`]: https://crates.io/crates/valuable
+//! [`format::Json`]: crate::fmt::format::Json
+//!
+//! ## Supported Rust Versions
+//!
+//! Tracing is built against the latest stable release. The minimum supported
+//! version is 1.65. The current Tracing version is not guaranteed to build on
+//! Rust versions earlier than the minimum supported version.
+//!
+//! Tracing follows the same compiler support policies as the rest of the Tokio
+//! project. The current stable Rust compiler and the three most recent minor
+//! versions before it will always be supported. For example, if the current
+//! stable compiler version is 1.69, the minimum supported version will not be
+//! increased past 1.66, three minor versions prior. Increasing the minimum
+//! supported compiler version is not considered a semver breaking change as
+//! long as doing so complies with this policy.
+//!
+//! [`Subscriber`]: tracing_core::subscriber::Subscriber
+//! [`tracing`]: https://docs.rs/tracing/latest/tracing
+//! [`EnvFilter`]: filter::EnvFilter
+//! [`fmt`]: mod@fmt
+//! [`tracing`]: https://crates.io/crates/tracing
+//! [`tracing-log`]: https://crates.io/crates/tracing-log
+//! [`smallvec`]: https://crates.io/crates/smallvec
+//! [`env_logger` crate]: https://crates.io/crates/env_logger
+//! [`parking_lot`]: https://crates.io/crates/parking_lot
+//! [`time` crate]: https://crates.io/crates/time
+//! [`liballoc`]: https://doc.rust-lang.org/alloc/index.html
+//! [`libstd`]: https://doc.rust-lang.org/std/index.html
+#![doc(
+    html_logo_url = "https://raw.githubusercontent.com/tokio-rs/tracing/main/assets/logo-type.png",
+    issue_tracker_base_url = "https://github.com/tokio-rs/tracing/issues/"
+)]
+#![cfg_attr(
+    docsrs,
+    // Allows displaying cfgs/feature flags in the documentation.
+    feature(doc_cfg),
+    // Allows adding traits to RustDoc's list of "notable traits"
+    feature(doc_notable_trait),
+    // Fail the docs build if any intra-docs links are broken
+    deny(rustdoc::broken_intra_doc_links),
+)]
+#![warn(
+    missing_debug_implementations,
+    missing_docs,
+    rust_2018_idioms,
+    unreachable_pub,
+    bad_style,
+    dead_code,
+    improper_ctypes,
+    non_shorthand_field_patterns,
+    no_mangle_generic_items,
+    overflowing_literals,
+    path_statements,
+    patterns_in_fns_without_body,
+    private_interfaces,
+    private_bounds,
+    unconditional_recursion,
+    unused,
+    unused_allocation,
+    unused_comparisons,
+    unused_parens,
+    while_true
+)]
+// Using struct update syntax when a struct has no additional fields avoids
+// a potential source change if additional fields are added to the struct in the
+// future, reducing diff noise. Allow this even though clippy considers it
+// "needless".
+#![allow(clippy::needless_update)]
+#![cfg_attr(not(feature = "std"), no_std)]
+
+#[cfg(feature = "alloc")]
+extern crate alloc;
+
+#[macro_use]
+mod macros;
+
+pub mod field;
+pub mod filter;
+pub mod prelude;
+pub mod registry;
+
+pub mod layer;
+pub mod util;
+
+feature! {
+    #![feature = "std"]
+    pub mod reload;
+    pub(crate) mod sync;
+}
+
+feature! {
+    #![all(feature = "fmt", feature = "std")]
+    pub mod fmt;
+    pub use fmt::fmt;
+    pub use fmt::Subscriber as FmtSubscriber;
+}
+
+feature! {
+    #![all(feature = "env-filter", feature = "std")]
+    pub use filter::EnvFilter;
+}
+
+pub use layer::Layer;
+
+feature! {
+    #![all(feature = "registry", feature = "std")]
+    pub use registry::Registry;
+
+    /// Returns a default [`Registry`].
+    pub fn registry() -> Registry {
+        Registry::default()
+    }
+}
+
+mod sealed {
+    pub trait Sealed<A = ()> {}
+}

vendor/tracing-subscriber/src/macros.rs

@@ -0,0 +1,28 @@
+#[cfg(feature = "std")]
+macro_rules! try_lock {
+    ($lock:expr) => {
+        try_lock!($lock, else return)
+    };
+    ($lock:expr, else $els:expr) => {
+        if let ::core::result::Result::Ok(l) = $lock {
+            l
+        } else if std::thread::panicking() {
+            $els
+        } else {
+            panic!("lock poisoned")
+        }
+    };
+}
+
+macro_rules! feature {
+    (
+        #![$meta:meta]
+        $($item:item)*
+    ) => {
+        $(
+            #[cfg($meta)]
+            #[cfg_attr(docsrs, doc(cfg($meta)))]
+            $item
+        )*
+    }
+}

vendor/tracing-subscriber/src/prelude.rs

@@ -0,0 +1,19 @@
+//! The `tracing-subscriber` prelude.
+//!
+//! This brings into scope a number of extension traits that define methods on
+//! types defined here and in other crates.
+
+pub use crate::field::{
+    MakeExt as __tracing_subscriber_field_MakeExt,
+    RecordFields as __tracing_subscriber_field_RecordFields,
+};
+pub use crate::layer::{
+    Layer as __tracing_subscriber_Layer, SubscriberExt as __tracing_subscriber_SubscriberExt,
+};
+
+pub use crate::util::SubscriberInitExt as _;
+
+feature! {
+    #![all(feature = "fmt", feature = "std")]
+    pub use crate::fmt::writer::MakeWriterExt as _;
+}

vendor/tracing-subscriber/src/registry/extensions.rs

@@ -0,0 +1,274 @@
+// taken from https://github.com/hyperium/http/blob/master/src/extensions.rs.
+
+use crate::sync::{RwLockReadGuard, RwLockWriteGuard};
+use std::{
+    any::{Any, TypeId},
+    collections::HashMap,
+    fmt,
+    hash::{BuildHasherDefault, Hasher},
+};
+
+#[allow(warnings)]
+type AnyMap = HashMap<TypeId, Box<dyn Any + Send + Sync>, BuildHasherDefault<IdHasher>>;
+
+/// With TypeIds as keys, there's no need to hash them. They are already hashes
+/// themselves, coming from the compiler. The IdHasher holds the u64 of
+/// the TypeId, and then returns it, instead of doing any bit fiddling.
+#[derive(Default, Debug)]
+struct IdHasher(u64);
+
+impl Hasher for IdHasher {
+    fn write(&mut self, _: &[u8]) {
+        unreachable!("TypeId calls write_u64");
+    }
+
+    #[inline]
+    fn write_u64(&mut self, id: u64) {
+        self.0 = id;
+    }
+
+    #[inline]
+    fn finish(&self) -> u64 {
+        self.0
+    }
+}
+
+/// An immutable, read-only reference to a Span's extensions.
+#[derive(Debug)]
+#[cfg_attr(docsrs, doc(cfg(feature = "std")))]
+pub struct Extensions<'a> {
+    inner: RwLockReadGuard<'a, ExtensionsInner>,
+}
+
+impl<'a> Extensions<'a> {
+    #[cfg(feature = "registry")]
+    pub(crate) fn new(inner: RwLockReadGuard<'a, ExtensionsInner>) -> Self {
+        Self { inner }
+    }
+
+    /// Immutably borrows a type previously inserted into this `Extensions`.
+    pub fn get<T: 'static>(&self) -> Option<&T> {
+        self.inner.get::<T>()
+    }
+}
+
+/// An mutable reference to a Span's extensions.
+#[derive(Debug)]
+#[cfg_attr(docsrs, doc(cfg(feature = "std")))]
+pub struct ExtensionsMut<'a> {
+    inner: RwLockWriteGuard<'a, ExtensionsInner>,
+}
+
+impl<'a> ExtensionsMut<'a> {
+    #[cfg(feature = "registry")]
+    pub(crate) fn new(inner: RwLockWriteGuard<'a, ExtensionsInner>) -> Self {
+        Self { inner }
+    }
+
+    /// Insert a type into this `Extensions`.
+    ///
+    /// Note that extensions are _not_
+    /// `Layer`-specific—they are _span_-specific. This means that
+    /// other layers can access and mutate extensions that
+    /// a different Layer recorded. For example, an application might
+    /// have a layer that records execution timings, alongside a layer
+    /// that reports spans and events to a distributed
+    /// tracing system that requires timestamps for spans.
+    /// Ideally, if one layer records a timestamp _x_, the other layer
+    /// should be able to reuse timestamp _x_.
+    ///
+    /// Therefore, extensions should generally be newtypes, rather than common
+    /// types like [`String`](std::string::String), to avoid accidental
+    /// cross-`Layer` clobbering.
+    ///
+    /// ## Panics
+    ///
+    /// If `T` is already present in `Extensions`, then this method will panic.
+    pub fn insert<T: Send + Sync + 'static>(&mut self, val: T) {
+        assert!(self.replace(val).is_none())
+    }
+
+    /// Replaces an existing `T` into this extensions.
+    ///
+    /// If `T` is not present, `Option::None` will be returned.
+    pub fn replace<T: Send + Sync + 'static>(&mut self, val: T) -> Option<T> {
+        self.inner.insert(val)
+    }
+
+    /// Get a mutable reference to a type previously inserted on this `ExtensionsMut`.
+    pub fn get_mut<T: 'static>(&mut self) -> Option<&mut T> {
+        self.inner.get_mut::<T>()
+    }
+
+    /// Remove a type from this `Extensions`.
+    ///
+    /// If a extension of this type existed, it will be returned.
+    pub fn remove<T: Send + Sync + 'static>(&mut self) -> Option<T> {
+        self.inner.remove::<T>()
+    }
+}
+
+/// A type map of span extensions.
+///
+/// [ExtensionsInner] is used by `SpanData` to store and
+/// span-specific data. A given `Layer` can read and write
+/// data that it is interested in recording and emitting.
+#[derive(Default)]
+pub(crate) struct ExtensionsInner {
+    map: AnyMap,
+}
+
+impl ExtensionsInner {
+    /// Create an empty `Extensions`.
+    #[cfg(any(test, feature = "registry"))]
+    #[inline]
+    #[cfg(any(test, feature = "registry"))]
+    pub(crate) fn new() -> ExtensionsInner {
+        ExtensionsInner {
+            map: AnyMap::default(),
+        }
+    }
+
+    /// Insert a type into this `Extensions`.
+    ///
+    /// If a extension of this type already existed, it will
+    /// be returned.
+    pub(crate) fn insert<T: Send + Sync + 'static>(&mut self, val: T) -> Option<T> {
+        self.map
+            .insert(TypeId::of::<T>(), Box::new(val))
+            .and_then(|boxed| {
+                #[allow(warnings)]
+                {
+                    (boxed as Box<Any + 'static>)
+                        .downcast()
+                        .ok()
+                        .map(|boxed| *boxed)
+                }
+            })
+    }
+
+    /// Get a reference to a type previously inserted on this `Extensions`.
+    pub(crate) fn get<T: 'static>(&self) -> Option<&T> {
+        self.map
+            .get(&TypeId::of::<T>())
+            .and_then(|boxed| (&**boxed as &(dyn Any + 'static)).downcast_ref())
+    }
+
+    /// Get a mutable reference to a type previously inserted on this `Extensions`.
+    pub(crate) fn get_mut<T: 'static>(&mut self) -> Option<&mut T> {
+        self.map
+            .get_mut(&TypeId::of::<T>())
+            .and_then(|boxed| (&mut **boxed as &mut (dyn Any + 'static)).downcast_mut())
+    }
+
+    /// Remove a type from this `Extensions`.
+    ///
+    /// If a extension of this type existed, it will be returned.
+    pub(crate) fn remove<T: Send + Sync + 'static>(&mut self) -> Option<T> {
+        self.map.remove(&TypeId::of::<T>()).and_then(|boxed| {
+            #[allow(warnings)]
+            {
+                (boxed as Box<Any + 'static>)
+                    .downcast()
+                    .ok()
+                    .map(|boxed| *boxed)
+            }
+        })
+    }
+
+    /// Clear the `ExtensionsInner` in-place, dropping any elements in the map but
+    /// retaining allocated capacity.
+    ///
+    /// This permits the hash map allocation to be pooled by the registry so
+    /// that future spans will not need to allocate new hashmaps.
+    #[cfg(any(test, feature = "registry"))]
+    pub(crate) fn clear(&mut self) {
+        self.map.clear();
+    }
+}
+
+impl fmt::Debug for ExtensionsInner {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        f.debug_struct("Extensions")
+            .field("len", &self.map.len())
+            .field("capacity", &self.map.capacity())
+            .finish()
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[derive(Debug, PartialEq)]
+    struct MyType(i32);
+
+    #[test]
+    fn test_extensions() {
+        let mut extensions = ExtensionsInner::new();
+
+        extensions.insert(5i32);
+        extensions.insert(MyType(10));
+
+        assert_eq!(extensions.get(), Some(&5i32));
+        assert_eq!(extensions.get_mut(), Some(&mut 5i32));
+
+        assert_eq!(extensions.remove::<i32>(), Some(5i32));
+        assert!(extensions.get::<i32>().is_none());
+
+        assert_eq!(extensions.get::<bool>(), None);
+        assert_eq!(extensions.get(), Some(&MyType(10)));
+    }
+
+    #[test]
+    fn clear_retains_capacity() {
+        let mut extensions = ExtensionsInner::new();
+        extensions.insert(5i32);
+        extensions.insert(MyType(10));
+        extensions.insert(true);
+
+        assert_eq!(extensions.map.len(), 3);
+        let prev_capacity = extensions.map.capacity();
+        extensions.clear();
+
+        assert_eq!(
+            extensions.map.len(),
+            0,
+            "after clear(), extensions map should have length 0"
+        );
+        assert_eq!(
+            extensions.map.capacity(),
+            prev_capacity,
+            "after clear(), extensions map should retain prior capacity"
+        );
+    }
+
+    #[test]
+    fn clear_drops_elements() {
+        use std::sync::Arc;
+        struct DropMePlease(Arc<()>);
+        struct DropMeTooPlease(Arc<()>);
+
+        let mut extensions = ExtensionsInner::new();
+        let val1 = DropMePlease(Arc::new(()));
+        let val2 = DropMeTooPlease(Arc::new(()));
+
+        let val1_dropped = Arc::downgrade(&val1.0);
+        let val2_dropped = Arc::downgrade(&val2.0);
+        extensions.insert(val1);
+        extensions.insert(val2);
+
+        assert!(val1_dropped.upgrade().is_some());
+        assert!(val2_dropped.upgrade().is_some());
+
+        extensions.clear();
+        assert!(
+            val1_dropped.upgrade().is_none(),
+            "after clear(), val1 should be dropped"
+        );
+        assert!(
+            val2_dropped.upgrade().is_none(),
+            "after clear(), val2 should be dropped"
+        );
+    }
+}

vendor/tracing-subscriber/src/registry/mod.rs

@@ -0,0 +1,598 @@
+//! Storage for span data shared by multiple [`Layer`]s.
+//!
+//! ## Using the Span Registry
+//!
+//! This module provides the [`Registry`] type, a [`Subscriber`] implementation
+//! which tracks per-span data and exposes it to [`Layer`]s. When a `Registry`
+//! is used as the base `Subscriber` of a `Layer` stack, the
+//! [`layer::Context`][ctx] type will provide methods allowing `Layer`s to
+//! [look up span data][lookup] stored in the registry. While [`Registry`] is a
+//! reasonable default for storing spans and events, other stores that implement
+//! [`LookupSpan`] and [`Subscriber`] themselves (with [`SpanData`] implemented
+//! by the per-span data they store) can be used as a drop-in replacement.
+//!
+//! For example, we might create a `Registry` and add multiple `Layer`s like so:
+//! ```rust
+//! use tracing_subscriber::{registry::Registry, Layer, prelude::*};
+//! # use tracing_core::Subscriber;
+//! # pub struct FooLayer {}
+//! # pub struct BarLayer {}
+//! # impl<S: Subscriber> Layer<S> for FooLayer {}
+//! # impl<S: Subscriber> Layer<S> for BarLayer {}
+//! # impl FooLayer {
+//! # fn new() -> Self { Self {} }
+//! # }
+//! # impl BarLayer {
+//! # fn new() -> Self { Self {} }
+//! # }
+//!
+//! let subscriber = Registry::default()
+//!     .with(FooLayer::new())
+//!     .with(BarLayer::new());
+//! ```
+//!
+//! If a type implementing `Layer` depends on the functionality of a `Registry`
+//! implementation, it should bound its `Subscriber` type parameter with the
+//! [`LookupSpan`] trait, like so:
+//!
+//! ```rust
+//! use tracing_subscriber::{registry, Layer};
+//! use tracing_core::Subscriber;
+//!
+//! pub struct MyLayer {
+//!     // ...
+//! }
+//!
+//! impl<S> Layer<S> for MyLayer
+//! where
+//!     S: Subscriber + for<'a> registry::LookupSpan<'a>,
+//! {
+//!     // ...
+//! }
+//! ```
+//! When this bound is added, the `Layer` implementation will be guaranteed
+//! access to the [`Context`][ctx] methods, such as [`Context::span`][lookup], that
+//! require the root subscriber to be a registry.
+//!
+//! [`Layer`]: crate::layer::Layer
+//! [`Subscriber`]: tracing_core::Subscriber
+//! [ctx]: crate::layer::Context
+//! [lookup]: crate::layer::Context::span()
+use tracing_core::{field::FieldSet, span::Id, Metadata};
+
+feature! {
+    #![feature = "std"]
+    /// A module containing a type map of span extensions.
+    mod extensions;
+    pub use extensions::{Extensions, ExtensionsMut};
+
+}
+
+feature! {
+    #![all(feature = "registry", feature = "std")]
+
+    mod sharded;
+    mod stack;
+
+    pub use sharded::Data;
+    pub use sharded::Registry;
+
+    use crate::filter::FilterId;
+}
+
+/// Provides access to stored span data.
+///
+/// Subscribers which store span data and associate it with span IDs should
+/// implement this trait; if they do, any [`Layer`]s wrapping them can look up
+/// metadata via the [`Context`] type's [`span()`] method.
+///
+/// [`Layer`]: super::layer::Layer
+/// [`Context`]: super::layer::Context
+/// [`span()`]: super::layer::Context::span
+pub trait LookupSpan<'a> {
+    /// The type of span data stored in this registry.
+    type Data: SpanData<'a>;
+
+    /// Returns the [`SpanData`] for a given `Id`, if it exists.
+    ///
+    /// <pre class="ignore" style="white-space:normal;font:inherit;">
+    /// <strong>Note</strong>: users of the <code>LookupSpan</code> trait should
+    /// typically call the <a href="#method.span"><code>span</code></a> method rather
+    /// than this method. The <code>span</code> method is implemented by
+    /// <em>calling</em> <code>span_data</code>, but returns a reference which is
+    /// capable of performing more sophisiticated queries.
+    /// </pre>
+    ///
+    fn span_data(&'a self, id: &Id) -> Option<Self::Data>;
+
+    /// Returns a [`SpanRef`] for the span with the given `Id`, if it exists.
+    ///
+    /// A `SpanRef` is similar to [`SpanData`], but it allows performing
+    /// additional lookups against the registryr that stores the wrapped data.
+    ///
+    /// In general, _users_ of the `LookupSpan` trait should use this method
+    /// rather than the [`span_data`] method; while _implementors_ of this trait
+    /// should only implement `span_data`.
+    ///
+    /// [`span_data`]: LookupSpan::span_data()
+    fn span(&'a self, id: &Id) -> Option<SpanRef<'a, Self>>
+    where
+        Self: Sized,
+    {
+        let data = self.span_data(id)?;
+        Some(SpanRef {
+            registry: self,
+            data,
+            #[cfg(feature = "registry")]
+            filter: FilterId::none(),
+        })
+    }
+
+    /// Registers a [`Filter`] for [per-layer filtering] with this
+    /// [`Subscriber`].
+    ///
+    /// The [`Filter`] can then use the returned [`FilterId`] to
+    /// [check if it previously enabled a span][check].
+    ///
+    /// # Panics
+    ///
+    /// If this `Subscriber` does not support [per-layer filtering].
+    ///
+    /// [`Filter`]: crate::layer::Filter
+    /// [per-layer filtering]: crate::layer::Layer#per-layer-filtering
+    /// [`Subscriber`]: tracing_core::Subscriber
+    /// [`FilterId`]: crate::filter::FilterId
+    /// [check]: SpanData::is_enabled_for
+    #[cfg(feature = "registry")]
+    #[cfg_attr(docsrs, doc(cfg(feature = "registry")))]
+    fn register_filter(&mut self) -> FilterId {
+        panic!(
+            "{} does not currently support filters",
+            std::any::type_name::<Self>()
+        )
+    }
+}
+
+/// A stored representation of data associated with a span.
+pub trait SpanData<'a> {
+    /// Returns this span's ID.
+    fn id(&self) -> Id;
+
+    /// Returns a reference to the span's `Metadata`.
+    fn metadata(&self) -> &'static Metadata<'static>;
+
+    /// Returns a reference to the ID
+    fn parent(&self) -> Option<&Id>;
+
+    /// Returns a reference to this span's `Extensions`.
+    ///
+    /// The extensions may be used by `Layer`s to store additional data
+    /// describing the span.
+    #[cfg(feature = "std")]
+    #[cfg_attr(docsrs, doc(cfg(feature = "std")))]
+    fn extensions(&self) -> Extensions<'_>;
+
+    /// Returns a mutable reference to this span's `Extensions`.
+    ///
+    /// The extensions may be used by `Layer`s to store additional data
+    /// describing the span.
+    #[cfg(feature = "std")]
+    #[cfg_attr(docsrs, doc(cfg(feature = "std")))]
+    fn extensions_mut(&self) -> ExtensionsMut<'_>;
+
+    /// Returns `true` if this span is enabled for the [per-layer filter][plf]
+    /// corresponding to the provided [`FilterId`].
+    ///
+    /// ## Default Implementation
+    ///
+    /// By default, this method assumes that the [`LookupSpan`] implementation
+    /// does not support [per-layer filtering][plf], and always returns `true`.
+    ///
+    /// [plf]: crate::layer::Layer#per-layer-filtering
+    /// [`FilterId`]: crate::filter::FilterId
+    #[cfg(feature = "registry")]
+    #[cfg_attr(docsrs, doc(cfg(feature = "registry")))]
+    fn is_enabled_for(&self, filter: FilterId) -> bool {
+        let _ = filter;
+        true
+    }
+}
+
+/// A reference to [span data] and the associated [registry].
+///
+/// This type implements all the same methods as [`SpanData`], and provides
+/// additional methods for querying the registry based on values from the span.
+///
+/// [registry]: LookupSpan
+#[derive(Debug)]
+pub struct SpanRef<'a, R: LookupSpan<'a>> {
+    registry: &'a R,
+    data: R::Data,
+
+    #[cfg(feature = "registry")]
+    filter: FilterId,
+}
+
+/// An iterator over the parents of a span, ordered from leaf to root.
+///
+/// This is returned by the [`SpanRef::scope`] method.
+#[derive(Debug)]
+pub struct Scope<'a, R> {
+    registry: &'a R,
+    next: Option<Id>,
+
+    #[cfg(all(feature = "registry", feature = "std"))]
+    filter: FilterId,
+}
+
+feature! {
+    #![any(feature = "alloc", feature = "std")]
+
+    #[cfg(not(feature = "smallvec"))]
+    use alloc::vec::{self, Vec};
+
+    use core::{fmt,iter};
+
+    /// An iterator over the parents of a span, ordered from root to leaf.
+    ///
+    /// This is returned by the [`Scope::from_root`] method.
+    pub struct ScopeFromRoot<'a, R>
+    where
+        R: LookupSpan<'a>,
+    {
+        #[cfg(feature = "smallvec")]
+        spans: iter::Rev<smallvec::IntoIter<SpanRefVecArray<'a, R>>>,
+        #[cfg(not(feature = "smallvec"))]
+        spans: iter::Rev<vec::IntoIter<SpanRef<'a, R>>>,
+    }
+
+    #[cfg(feature = "smallvec")]
+    type SpanRefVecArray<'span, L> = [SpanRef<'span, L>; 16];
+
+    impl<'a, R> Scope<'a, R>
+    where
+        R: LookupSpan<'a>,
+    {
+        /// Flips the order of the iterator, so that it is ordered from root to leaf.
+        ///
+        /// The iterator will first return the root span, then that span's immediate child,
+        /// and so on until it finally returns the span that [`SpanRef::scope`] was called on.
+        ///
+        /// If any items were consumed from the [`Scope`] before calling this method then they
+        /// will *not* be returned from the [`ScopeFromRoot`].
+        ///
+        /// **Note**: this will allocate if there are many spans remaining, or if the
+        /// "smallvec" feature flag is not enabled.
+        #[allow(clippy::wrong_self_convention)]
+        pub fn from_root(self) -> ScopeFromRoot<'a, R> {
+            #[cfg(feature = "smallvec")]
+            type Buf<T> = smallvec::SmallVec<T>;
+            #[cfg(not(feature = "smallvec"))]
+            type Buf<T> = Vec<T>;
+            ScopeFromRoot {
+                spans: self.collect::<Buf<_>>().into_iter().rev(),
+            }
+        }
+    }
+
+    impl<'a, R> Iterator for ScopeFromRoot<'a, R>
+    where
+        R: LookupSpan<'a>,
+    {
+        type Item = SpanRef<'a, R>;
+
+        #[inline]
+        fn next(&mut self) -> Option<Self::Item> {
+            self.spans.next()
+        }
+
+        #[inline]
+        fn size_hint(&self) -> (usize, Option<usize>) {
+            self.spans.size_hint()
+        }
+    }
+
+    impl<'a, R> fmt::Debug for ScopeFromRoot<'a, R>
+    where
+        R: LookupSpan<'a>,
+    {
+        fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+            f.pad("ScopeFromRoot { .. }")
+        }
+    }
+}
+
+impl<'a, R> Iterator for Scope<'a, R>
+where
+    R: LookupSpan<'a>,
+{
+    type Item = SpanRef<'a, R>;
+
+    fn next(&mut self) -> Option<Self::Item> {
+        loop {
+            let curr = self.registry.span(self.next.as_ref()?)?;
+
+            #[cfg(all(feature = "registry", feature = "std"))]
+            let curr = curr.with_filter(self.filter);
+            self.next = curr.data.parent().cloned();
+
+            // If the `Scope` is filtered, check if the current span is enabled
+            // by the selected filter ID.
+
+            #[cfg(all(feature = "registry", feature = "std"))]
+            {
+                if !curr.is_enabled_for(self.filter) {
+                    // The current span in the chain is disabled for this
+                    // filter. Try its parent.
+                    continue;
+                }
+            }
+
+            return Some(curr);
+        }
+    }
+}
+
+impl<'a, R> SpanRef<'a, R>
+where
+    R: LookupSpan<'a>,
+{
+    /// Returns this span's ID.
+    pub fn id(&self) -> Id {
+        self.data.id()
+    }
+
+    /// Returns a static reference to the span's metadata.
+    pub fn metadata(&self) -> &'static Metadata<'static> {
+        self.data.metadata()
+    }
+
+    /// Returns the span's name,
+    pub fn name(&self) -> &'static str {
+        self.data.metadata().name()
+    }
+
+    /// Returns a list of [fields] defined by the span.
+    ///
+    /// [fields]: tracing_core::field
+    pub fn fields(&self) -> &FieldSet {
+        self.data.metadata().fields()
+    }
+
+    /// Returns a `SpanRef` describing this span's parent, or `None` if this
+    /// span is the root of its trace tree.
+    pub fn parent(&self) -> Option<Self> {
+        let id = self.data.parent()?;
+        let data = self.registry.span_data(id)?;
+
+        #[cfg(all(feature = "registry", feature = "std"))]
+        {
+            // move these into mut bindings if the registry feature is enabled,
+            // since they may be mutated in the loop.
+            let mut data = data;
+            loop {
+                // Is this parent enabled by our filter?
+                if data.is_enabled_for(self.filter) {
+                    return Some(Self {
+                        registry: self.registry,
+                        filter: self.filter,
+                        data,
+                    });
+                }
+
+                // It's not enabled. If the disabled span has a parent, try that!
+                let id = data.parent()?;
+                data = self.registry.span_data(id)?;
+            }
+        }
+
+        #[cfg(not(all(feature = "registry", feature = "std")))]
+        Some(Self {
+            registry: self.registry,
+            data,
+        })
+    }
+
+    /// Returns an iterator over all parents of this span, starting with this span,
+    /// ordered from leaf to root.
+    ///
+    /// The iterator will first return the span, then the span's immediate parent,
+    /// followed by that span's parent, and so on, until it reaches a root span.
+    ///
+    /// ```rust
+    /// use tracing::{span, Subscriber};
+    /// use tracing_subscriber::{
+    ///     layer::{Context, Layer},
+    ///     prelude::*,
+    ///     registry::LookupSpan,
+    /// };
+    ///
+    /// struct PrintingLayer;
+    /// impl<S> Layer<S> for PrintingLayer
+    /// where
+    ///     S: Subscriber + for<'lookup> LookupSpan<'lookup>,
+    /// {
+    ///     fn on_enter(&self, id: &span::Id, ctx: Context<S>) {
+    ///         let span = ctx.span(id).unwrap();
+    ///         let scope = span.scope().map(|span| span.name()).collect::<Vec<_>>();
+    ///         println!("Entering span: {:?}", scope);
+    ///     }
+    /// }
+    ///
+    /// tracing::subscriber::with_default(tracing_subscriber::registry().with(PrintingLayer), || {
+    ///     let _root = tracing::info_span!("root").entered();
+    ///     // Prints: Entering span: ["root"]
+    ///     let _child = tracing::info_span!("child").entered();
+    ///     // Prints: Entering span: ["child", "root"]
+    ///     let _leaf = tracing::info_span!("leaf").entered();
+    ///     // Prints: Entering span: ["leaf", "child", "root"]
+    /// });
+    /// ```
+    ///
+    /// If the opposite order (from the root to this span) is desired, calling [`Scope::from_root`] on
+    /// the returned iterator reverses the order.
+    ///
+    /// ```rust
+    /// # use tracing::{span, Subscriber};
+    /// # use tracing_subscriber::{
+    /// #     layer::{Context, Layer},
+    /// #     prelude::*,
+    /// #     registry::LookupSpan,
+    /// # };
+    /// # struct PrintingLayer;
+    /// impl<S> Layer<S> for PrintingLayer
+    /// where
+    ///     S: Subscriber + for<'lookup> LookupSpan<'lookup>,
+    /// {
+    ///     fn on_enter(&self, id: &span::Id, ctx: Context<S>) {
+    ///         let span = ctx.span(id).unwrap();
+    ///         let scope = span.scope().from_root().map(|span| span.name()).collect::<Vec<_>>();
+    ///         println!("Entering span: {:?}", scope);
+    ///     }
+    /// }
+    ///
+    /// tracing::subscriber::with_default(tracing_subscriber::registry().with(PrintingLayer), || {
+    ///     let _root = tracing::info_span!("root").entered();
+    ///     // Prints: Entering span: ["root"]
+    ///     let _child = tracing::info_span!("child").entered();
+    ///     // Prints: Entering span: ["root", "child"]
+    ///     let _leaf = tracing::info_span!("leaf").entered();
+    ///     // Prints: Entering span: ["root", "child", "leaf"]
+    /// });
+    /// ```
+    pub fn scope(&self) -> Scope<'a, R> {
+        Scope {
+            registry: self.registry,
+            next: Some(self.id()),
+
+            #[cfg(feature = "registry")]
+            filter: self.filter,
+        }
+    }
+
+    /// Returns a reference to this span's `Extensions`.
+    ///
+    /// The extensions may be used by `Layer`s to store additional data
+    /// describing the span.
+    #[cfg(feature = "std")]
+    #[cfg_attr(docsrs, doc(cfg(feature = "std")))]
+    pub fn extensions(&self) -> Extensions<'_> {
+        self.data.extensions()
+    }
+
+    /// Returns a mutable reference to this span's `Extensions`.
+    ///
+    /// The extensions may be used by `Layer`s to store additional data
+    /// describing the span.
+    #[cfg(feature = "std")]
+    #[cfg_attr(docsrs, doc(cfg(feature = "std")))]
+    pub fn extensions_mut(&self) -> ExtensionsMut<'_> {
+        self.data.extensions_mut()
+    }
+
+    #[cfg(all(feature = "registry", feature = "std"))]
+    pub(crate) fn try_with_filter(self, filter: FilterId) -> Option<Self> {
+        if self.is_enabled_for(filter) {
+            return Some(self.with_filter(filter));
+        }
+
+        None
+    }
+
+    #[inline]
+    #[cfg(all(feature = "registry", feature = "std"))]
+    pub(crate) fn is_enabled_for(&self, filter: FilterId) -> bool {
+        self.data.is_enabled_for(filter)
+    }
+
+    #[inline]
+    #[cfg(all(feature = "registry", feature = "std"))]
+    fn with_filter(self, filter: FilterId) -> Self {
+        Self { filter, ..self }
+    }
+}
+
+#[cfg(all(test, feature = "registry", feature = "std"))]
+mod tests {
+    use crate::{
+        layer::{Context, Layer},
+        prelude::*,
+        registry::LookupSpan,
+    };
+    use std::sync::{Arc, Mutex};
+    use tracing::{span, Subscriber};
+
+    #[test]
+    fn spanref_scope_iteration_order() {
+        let last_entered_scope = Arc::new(Mutex::new(Vec::new()));
+
+        #[derive(Default)]
+        struct PrintingLayer {
+            last_entered_scope: Arc<Mutex<Vec<&'static str>>>,
+        }
+
+        impl<S> Layer<S> for PrintingLayer
+        where
+            S: Subscriber + for<'lookup> LookupSpan<'lookup>,
+        {
+            fn on_enter(&self, id: &span::Id, ctx: Context<'_, S>) {
+                let span = ctx.span(id).unwrap();
+                let scope = span.scope().map(|span| span.name()).collect::<Vec<_>>();
+                *self.last_entered_scope.lock().unwrap() = scope;
+            }
+        }
+
+        let _guard = tracing::subscriber::set_default(crate::registry().with(PrintingLayer {
+            last_entered_scope: last_entered_scope.clone(),
+        }));
+
+        let _root = tracing::info_span!("root").entered();
+        assert_eq!(&*last_entered_scope.lock().unwrap(), &["root"]);
+        let _child = tracing::info_span!("child").entered();
+        assert_eq!(&*last_entered_scope.lock().unwrap(), &["child", "root"]);
+        let _leaf = tracing::info_span!("leaf").entered();
+        assert_eq!(
+            &*last_entered_scope.lock().unwrap(),
+            &["leaf", "child", "root"]
+        );
+    }
+
+    #[test]
+    fn spanref_scope_fromroot_iteration_order() {
+        let last_entered_scope = Arc::new(Mutex::new(Vec::new()));
+
+        #[derive(Default)]
+        struct PrintingLayer {
+            last_entered_scope: Arc<Mutex<Vec<&'static str>>>,
+        }
+
+        impl<S> Layer<S> for PrintingLayer
+        where
+            S: Subscriber + for<'lookup> LookupSpan<'lookup>,
+        {
+            fn on_enter(&self, id: &span::Id, ctx: Context<'_, S>) {
+                let span = ctx.span(id).unwrap();
+                let scope = span
+                    .scope()
+                    .from_root()
+                    .map(|span| span.name())
+                    .collect::<Vec<_>>();
+                *self.last_entered_scope.lock().unwrap() = scope;
+            }
+        }
+
+        let _guard = tracing::subscriber::set_default(crate::registry().with(PrintingLayer {
+            last_entered_scope: last_entered_scope.clone(),
+        }));
+
+        let _root = tracing::info_span!("root").entered();
+        assert_eq!(&*last_entered_scope.lock().unwrap(), &["root"]);
+        let _child = tracing::info_span!("child").entered();
+        assert_eq!(&*last_entered_scope.lock().unwrap(), &["root", "child",]);
+        let _leaf = tracing::info_span!("leaf").entered();
+        assert_eq!(
+            &*last_entered_scope.lock().unwrap(),
+            &["root", "child", "leaf"]
+        );
+    }
+}

vendor/tracing-subscriber/src/registry/sharded.rs

@@ -0,0 +1,905 @@
+use sharded_slab::{pool::Ref, Clear, Pool};
+use thread_local::ThreadLocal;
+
+use super::stack::SpanStack;
+use crate::{
+    filter::{FilterId, FilterMap, FilterState},
+    registry::{
+        extensions::{Extensions, ExtensionsInner, ExtensionsMut},
+        LookupSpan, SpanData,
+    },
+    sync::RwLock,
+};
+use std::{
+    cell::{self, Cell, RefCell},
+    sync::atomic::{fence, AtomicUsize, Ordering},
+};
+use tracing_core::{
+    dispatcher::{self, Dispatch},
+    span::{self, Current, Id},
+    Event, Interest, Metadata, Subscriber,
+};
+
+/// A shared, reusable store for spans.
+///
+/// A `Registry` is a [`Subscriber`] around which multiple [`Layer`]s
+/// implementing various behaviors may be [added]. Unlike other types
+/// implementing `Subscriber`, `Registry` does not actually record traces itself:
+/// instead, it collects and stores span data that is exposed to any [`Layer`]s
+/// wrapping it through implementations of the [`LookupSpan`] trait.
+/// The `Registry` is responsible for storing span metadata, recording
+/// relationships between spans, and tracking which spans are active and which
+/// are closed. In addition, it provides a mechanism for [`Layer`]s to store
+/// user-defined per-span data, called [extensions], in the registry. This
+/// allows [`Layer`]-specific data to benefit from the `Registry`'s
+/// high-performance concurrent storage.
+///
+/// This registry is implemented using a [lock-free sharded slab][slab], and is
+/// highly optimized for concurrent access.
+///
+/// # Span ID Generation
+///
+/// Span IDs are not globally unique, but the registry ensures that
+/// no two currently active spans have the same ID within a process.
+///
+/// One of the primary responsibilities of the registry is to generate [span
+/// IDs]. Therefore, it's important for other code that interacts with the
+/// registry, such as [`Layer`]s, to understand the guarantees of the
+/// span IDs that are generated.
+///
+/// The registry's span IDs are guaranteed to be unique **at a given point
+/// in time**. This means that an active span will never be assigned the
+/// same ID as another **currently active** span. However, the registry
+/// **will** eventually reuse the IDs of [closed] spans, although an ID
+/// will never be reassigned immediately after a span has closed.
+///
+/// Spans are not [considered closed] by the `Registry` until *every*
+/// [`Span`] reference with that ID has been dropped.
+///
+/// Thus: span IDs generated by the registry should be considered unique
+/// only at a given point in time, and only relative to other spans
+/// generated by the same process. Two spans with the same ID will not exist
+/// in the same process concurrently. However, if historical span data is
+/// being stored, the same ID may occur for multiple spans times in that
+/// data. If spans must be uniquely identified in historical data, the user
+/// code storing this data must assign its own unique identifiers to those
+/// spans. A counter is generally sufficient for this.
+///
+/// Similarly, span IDs generated by the registry are not unique outside of
+/// a given process. Distributed tracing systems may require identifiers
+/// that are unique across multiple processes on multiple machines (for
+/// example, [OpenTelemetry's `SpanId`s and `TraceId`s][ot]). `tracing` span
+/// IDs generated by the registry should **not** be used for this purpose.
+/// Instead, code which integrates with a distributed tracing system should
+/// generate and propagate its own IDs according to the rules specified by
+/// the distributed tracing system. These IDs can be associated with
+/// `tracing` spans using [fields] and/or [stored span data].
+///
+/// [span IDs]: tracing_core::span::Id
+/// [slab]: sharded_slab
+/// [`Layer`]: crate::Layer
+/// [added]: crate::layer::Layer#composing-layers
+/// [extensions]: super::Extensions
+/// [closed]: https://docs.rs/tracing/latest/tracing/span/index.html#closing-spans
+/// [considered closed]: tracing_core::subscriber::Subscriber::try_close()
+/// [`Span`]: https://docs.rs/tracing/latest/tracing/span/struct.Span.html
+/// [ot]: https://github.com/open-telemetry/opentelemetry-specification/blob/main/specification/trace/api.md#spancontext
+/// [fields]: tracing_core::field
+/// [stored span data]: crate::registry::SpanData::extensions_mut
+#[cfg(feature = "registry")]
+#[cfg_attr(docsrs, doc(cfg(all(feature = "registry", feature = "std"))))]
+#[derive(Debug)]
+pub struct Registry {
+    spans: Pool<DataInner>,
+    current_spans: ThreadLocal<RefCell<SpanStack>>,
+    next_filter_id: u8,
+}
+
+/// Span data stored in a [`Registry`].
+///
+/// The registry stores well-known data defined by tracing: span relationships,
+/// metadata and reference counts. Additional user-defined data provided by
+/// [`Layer`s], such as formatted fields, metrics, or distributed traces should
+/// be stored in the [extensions] typemap.
+///
+/// [`Layer`s]: crate::layer::Layer
+/// [extensions]: Extensions
+#[cfg(feature = "registry")]
+#[cfg_attr(docsrs, doc(cfg(all(feature = "registry", feature = "std"))))]
+#[derive(Debug)]
+pub struct Data<'a> {
+    /// Immutable reference to the pooled `DataInner` entry.
+    inner: Ref<'a, DataInner>,
+}
+
+/// Stored data associated with a span.
+///
+/// This type is pooled using [`sharded_slab::Pool`]; when a span is
+/// dropped, the `DataInner` entry at that span's slab index is cleared
+/// in place and reused by a future span. Thus, the `Default` and
+/// [`sharded_slab::Clear`] implementations for this type are
+/// load-bearing.
+#[derive(Debug)]
+struct DataInner {
+    filter_map: FilterMap,
+    metadata: &'static Metadata<'static>,
+    parent: Option<Id>,
+    ref_count: AtomicUsize,
+    // The span's `Extensions` typemap. Allocations for the `HashMap` backing
+    // this are pooled and reused in place.
+    pub(crate) extensions: RwLock<ExtensionsInner>,
+}
+
+// === impl Registry ===
+
+impl Default for Registry {
+    fn default() -> Self {
+        Self {
+            spans: Pool::new(),
+            current_spans: ThreadLocal::new(),
+            next_filter_id: 0,
+        }
+    }
+}
+
+#[inline]
+fn idx_to_id(idx: usize) -> Id {
+    Id::from_u64(idx as u64 + 1)
+}
+
+#[inline]
+fn id_to_idx(id: &Id) -> usize {
+    id.into_u64() as usize - 1
+}
+
+/// A guard that tracks how many [`Registry`]-backed `Layer`s have
+/// processed an `on_close` event.
+///
+/// This is needed to enable a [`Registry`]-backed Layer to access span
+/// data after the `Layer` has recieved the `on_close` callback.
+///
+/// Once all `Layer`s have processed this event, the [`Registry`] knows
+/// that is able to safely remove the span tracked by `id`. `CloseGuard`
+/// accomplishes this through a two-step process:
+/// 1. Whenever a [`Registry`]-backed `Layer::on_close` method is
+///    called, `Registry::start_close` is closed.
+///    `Registry::start_close` increments a thread-local `CLOSE_COUNT`
+///    by 1 and returns a `CloseGuard`.
+/// 2. The `CloseGuard` is dropped at the end of `Layer::on_close`. On
+///    drop, `CloseGuard` checks thread-local `CLOSE_COUNT`. If
+///    `CLOSE_COUNT` is 0, the `CloseGuard` removes the span with the
+///    `id` from the registry, as all `Layers` that might have seen the
+///    `on_close` notification have processed it. If `CLOSE_COUNT` is
+///    greater than 0, `CloseGuard` decrements the counter by one and
+///    _does not_ remove the span from the [`Registry`].
+///
+pub(crate) struct CloseGuard<'a> {
+    id: Id,
+    registry: &'a Registry,
+    is_closing: bool,
+}
+
+impl Registry {
+    fn get(&self, id: &Id) -> Option<Ref<'_, DataInner>> {
+        self.spans.get(id_to_idx(id))
+    }
+
+    /// Returns a guard which tracks how many `Layer`s have
+    /// processed an `on_close` notification via the `CLOSE_COUNT` thread-local.
+    /// For additional details, see [`CloseGuard`].
+    ///
+    pub(crate) fn start_close(&self, id: Id) -> CloseGuard<'_> {
+        CLOSE_COUNT.with(|count| {
+            let c = count.get();
+            count.set(c + 1);
+        });
+        CloseGuard {
+            id,
+            registry: self,
+            is_closing: false,
+        }
+    }
+
+    pub(crate) fn has_per_layer_filters(&self) -> bool {
+        self.next_filter_id > 0
+    }
+
+    pub(crate) fn span_stack(&self) -> cell::Ref<'_, SpanStack> {
+        self.current_spans.get_or_default().borrow()
+    }
+}
+
+thread_local! {
+    /// `CLOSE_COUNT` is the thread-local counter used by `CloseGuard` to
+    /// track how many layers have processed the close.
+    /// For additional details, see [`CloseGuard`].
+    ///
+    static CLOSE_COUNT: Cell<usize> = const { Cell::new(0) };
+}
+
+impl Subscriber for Registry {
+    fn register_callsite(&self, _: &'static Metadata<'static>) -> Interest {
+        if self.has_per_layer_filters() {
+            return FilterState::take_interest().unwrap_or_else(Interest::always);
+        }
+
+        Interest::always()
+    }
+
+    fn enabled(&self, _: &Metadata<'_>) -> bool {
+        if self.has_per_layer_filters() {
+            return FilterState::event_enabled();
+        }
+        true
+    }
+
+    #[inline]
+    fn new_span(&self, attrs: &span::Attributes<'_>) -> span::Id {
+        let parent = if attrs.is_root() {
+            None
+        } else if attrs.is_contextual() {
+            self.current_span().id().map(|id| self.clone_span(id))
+        } else {
+            attrs.parent().map(|id| self.clone_span(id))
+        };
+
+        let id = self
+            .spans
+            // Check out a `DataInner` entry from the pool for the new span. If
+            // there are free entries already allocated in the pool, this will
+            // preferentially reuse one; otherwise, a new `DataInner` is
+            // allocated and added to the pool.
+            .create_with(|data| {
+                data.metadata = attrs.metadata();
+                data.parent = parent;
+                data.filter_map = crate::filter::FILTERING.with(|filtering| filtering.filter_map());
+                #[cfg(debug_assertions)]
+                {
+                    if data.filter_map != FilterMap::new() {
+                        debug_assert!(self.has_per_layer_filters());
+                    }
+                }
+
+                let refs = data.ref_count.get_mut();
+                debug_assert_eq!(*refs, 0);
+                *refs = 1;
+            })
+            .expect("Unable to allocate another span");
+        idx_to_id(id)
+    }
+
+    /// This is intentionally not implemented, as recording fields
+    /// on a span is the responsibility of layers atop of this registry.
+    #[inline]
+    fn record(&self, _: &span::Id, _: &span::Record<'_>) {}
+
+    fn record_follows_from(&self, _span: &span::Id, _follows: &span::Id) {}
+
+    fn event_enabled(&self, _event: &Event<'_>) -> bool {
+        if self.has_per_layer_filters() {
+            return FilterState::event_enabled();
+        }
+        true
+    }
+
+    /// This is intentionally not implemented, as recording events
+    /// is the responsibility of layers atop of this registry.
+    fn event(&self, _: &Event<'_>) {}
+
+    fn enter(&self, id: &span::Id) {
+        if self
+            .current_spans
+            .get_or_default()
+            .borrow_mut()
+            .push(id.clone())
+        {
+            self.clone_span(id);
+        }
+    }
+
+    fn exit(&self, id: &span::Id) {
+        if let Some(spans) = self.current_spans.get() {
+            if spans.borrow_mut().pop(id) {
+                dispatcher::get_default(|dispatch| dispatch.try_close(id.clone()));
+            }
+        }
+    }
+
+    fn clone_span(&self, id: &span::Id) -> span::Id {
+        let span = self
+            .get(id)
+            .unwrap_or_else(|| panic!(
+                "tried to clone {:?}, but no span exists with that ID\n\
+                This may be caused by consuming a parent span (`parent: span`) rather than borrowing it (`parent: &span`).",
+                id,
+            ));
+        // Like `std::sync::Arc`, adds to the ref count (on clone) don't require
+        // a strong ordering; if we call` clone_span`, the reference count must
+        // always at least 1. The only synchronization necessary is between
+        // calls to `try_close`: we have to ensure that all threads have
+        // dropped their refs to the span before the span is closed.
+        let refs = span.ref_count.fetch_add(1, Ordering::Relaxed);
+        assert_ne!(
+            refs, 0,
+            "tried to clone a span ({:?}) that already closed",
+            id
+        );
+        id.clone()
+    }
+
+    fn current_span(&self) -> Current {
+        self.current_spans
+            .get()
+            .and_then(|spans| {
+                let spans = spans.borrow();
+                let id = spans.current()?;
+                let span = self.get(id)?;
+                Some(Current::new(id.clone(), span.metadata))
+            })
+            .unwrap_or_else(Current::none)
+    }
+
+    /// Decrements the reference count of the span with the given `id`, and
+    /// removes the span if it is zero.
+    ///
+    /// The allocated span slot will be reused when a new span is created.
+    fn try_close(&self, id: span::Id) -> bool {
+        let span = match self.get(&id) {
+            Some(span) => span,
+            None if std::thread::panicking() => return false,
+            None => panic!("tried to drop a ref to {:?}, but no such span exists!", id),
+        };
+
+        let refs = span.ref_count.fetch_sub(1, Ordering::Release);
+        if !std::thread::panicking() {
+            assert!(refs < usize::MAX, "reference count overflow!");
+        }
+        if refs > 1 {
+            return false;
+        }
+
+        // Synchronize if we are actually removing the span (stolen
+        // from std::Arc); this ensures that all other `try_close` calls on
+        // other threads happen-before we actually remove the span.
+        fence(Ordering::Acquire);
+        true
+    }
+}
+
+impl<'a> LookupSpan<'a> for Registry {
+    type Data = Data<'a>;
+
+    fn span_data(&'a self, id: &Id) -> Option<Self::Data> {
+        let inner = self.get(id)?;
+        Some(Data { inner })
+    }
+
+    fn register_filter(&mut self) -> FilterId {
+        let id = FilterId::new(self.next_filter_id);
+        self.next_filter_id += 1;
+        id
+    }
+}
+
+// === impl CloseGuard ===
+
+impl CloseGuard<'_> {
+    pub(crate) fn set_closing(&mut self) {
+        self.is_closing = true;
+    }
+}
+
+impl Drop for CloseGuard<'_> {
+    fn drop(&mut self) {
+        // If this returns with an error, we are already panicking. At
+        // this point, there's nothing we can really do to recover
+        // except by avoiding a double-panic.
+        let _ = CLOSE_COUNT.try_with(|count| {
+            let c = count.get();
+            // Decrement the count to indicate that _this_ guard's
+            // `on_close` callback has completed.
+            //
+            // Note that we *must* do this before we actually remove the span
+            // from the registry, since dropping the `DataInner` may trigger a
+            // new close, if this span is the last reference to a parent span.
+            count.set(c - 1);
+
+            // If the current close count is 1, this stack frame is the last
+            // `on_close` call. If the span is closing, it's okay to remove the
+            // span.
+            if c == 1 && self.is_closing {
+                self.registry.spans.clear(id_to_idx(&self.id));
+            }
+        });
+    }
+}
+
+// === impl Data ===
+
+impl<'a> SpanData<'a> for Data<'a> {
+    fn id(&self) -> Id {
+        idx_to_id(self.inner.key())
+    }
+
+    fn metadata(&self) -> &'static Metadata<'static> {
+        self.inner.metadata
+    }
+
+    fn parent(&self) -> Option<&Id> {
+        self.inner.parent.as_ref()
+    }
+
+    fn extensions(&self) -> Extensions<'_> {
+        Extensions::new(self.inner.extensions.read().expect("Mutex poisoned"))
+    }
+
+    fn extensions_mut(&self) -> ExtensionsMut<'_> {
+        ExtensionsMut::new(self.inner.extensions.write().expect("Mutex poisoned"))
+    }
+
+    #[inline]
+    fn is_enabled_for(&self, filter: FilterId) -> bool {
+        self.inner.filter_map.is_enabled(filter)
+    }
+}
+
+// === impl DataInner ===
+
+impl Default for DataInner {
+    fn default() -> Self {
+        // Since `DataInner` owns a `&'static Callsite` pointer, we need
+        // something to use as the initial default value for that callsite.
+        // Since we can't access a `DataInner` until it has had actual span data
+        // inserted into it, the null metadata will never actually be accessed.
+        struct NullCallsite;
+        impl tracing_core::callsite::Callsite for NullCallsite {
+            fn set_interest(&self, _: Interest) {
+                unreachable!(
+                    "/!\\ Tried to register the null callsite /!\\\n \
+                    This should never have happened and is definitely a bug. \
+                    A `tracing` bug report would be appreciated."
+                )
+            }
+
+            fn metadata(&self) -> &Metadata<'_> {
+                unreachable!(
+                    "/!\\ Tried to access the null callsite's metadata /!\\\n \
+                    This should never have happened and is definitely a bug. \
+                    A `tracing` bug report would be appreciated."
+                )
+            }
+        }
+
+        static NULL_CALLSITE: NullCallsite = NullCallsite;
+        static NULL_METADATA: Metadata<'static> = tracing_core::metadata! {
+            name: "",
+            target: "",
+            level: tracing_core::Level::TRACE,
+            fields: &[],
+            callsite: &NULL_CALLSITE,
+            kind: tracing_core::metadata::Kind::SPAN,
+        };
+
+        Self {
+            filter_map: FilterMap::new(),
+            metadata: &NULL_METADATA,
+            parent: None,
+            ref_count: AtomicUsize::new(0),
+            extensions: RwLock::new(ExtensionsInner::new()),
+        }
+    }
+}
+
+impl Clear for DataInner {
+    /// Clears the span's data in place, dropping the parent's reference count.
+    fn clear(&mut self) {
+        // A span is not considered closed until all of its children have closed.
+        // Therefore, each span's `DataInner` holds a "reference" to the parent
+        // span, keeping the parent span open until all its children have closed.
+        // When we close a span, we must then decrement the parent's ref count
+        // (potentially, allowing it to close, if this child is the last reference
+        // to that span).
+        // We have to actually unpack the option inside the `get_default`
+        // closure, since it is a `FnMut`, but testing that there _is_ a value
+        // here lets us avoid the thread-local access if we don't need the
+        // dispatcher at all.
+        if self.parent.is_some() {
+            // Note that --- because `Layered::try_close` works by calling
+            // `try_close` on the inner subscriber and using the return value to
+            // determine whether to call the `Layer`'s `on_close` callback ---
+            // we must call `try_close` on the entire subscriber stack, rather
+            // than just on the registry. If the registry called `try_close` on
+            // itself directly, the layers wouldn't see the close notification.
+            let subscriber = dispatcher::get_default(Dispatch::clone);
+            if let Some(parent) = self.parent.take() {
+                let _ = subscriber.try_close(parent);
+            }
+        }
+
+        // Clear (but do not deallocate!) the pooled `HashMap` for the span's extensions.
+        self.extensions
+            .get_mut()
+            .unwrap_or_else(|l| {
+                // This function can be called in a `Drop` impl, such as while
+                // panicking, so ignore lock poisoning.
+                l.into_inner()
+            })
+            .clear();
+
+        self.filter_map = FilterMap::new();
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use crate::{layer::Context, registry::LookupSpan, Layer};
+    use std::{
+        collections::HashMap,
+        sync::{Arc, Mutex, Weak},
+    };
+    use tracing::{self, subscriber::with_default};
+    use tracing_core::{
+        dispatcher,
+        span::{Attributes, Id},
+        Subscriber,
+    };
+
+    struct AssertionLayer;
+    impl<S> Layer<S> for AssertionLayer
+    where
+        S: Subscriber + for<'a> LookupSpan<'a>,
+    {
+        fn on_close(&self, id: Id, ctx: Context<'_, S>) {
+            dbg!(format_args!("closing {:?}", id));
+            assert!(&ctx.span(&id).is_some());
+        }
+    }
+
+    #[test]
+    fn single_layer_can_access_closed_span() {
+        let subscriber = AssertionLayer.with_subscriber(Registry::default());
+
+        with_default(subscriber, || {
+            let span = tracing::debug_span!("span");
+            drop(span);
+        });
+    }
+
+    #[test]
+    fn multiple_layers_can_access_closed_span() {
+        let subscriber = AssertionLayer
+            .and_then(AssertionLayer)
+            .with_subscriber(Registry::default());
+
+        with_default(subscriber, || {
+            let span = tracing::debug_span!("span");
+            drop(span);
+        });
+    }
+
+    struct CloseLayer {
+        inner: Arc<Mutex<CloseState>>,
+    }
+
+    struct CloseHandle {
+        state: Arc<Mutex<CloseState>>,
+    }
+
+    #[derive(Default)]
+    struct CloseState {
+        open: HashMap<&'static str, Weak<()>>,
+        closed: Vec<(&'static str, Weak<()>)>,
+    }
+
+    #[allow(dead_code)] // Field is exercised via checking `Arc::downgrade()`
+    struct SetRemoved(Arc<()>);
+
+    impl<S> Layer<S> for CloseLayer
+    where
+        S: Subscriber + for<'a> LookupSpan<'a>,
+    {
+        fn on_new_span(&self, _: &Attributes<'_>, id: &Id, ctx: Context<'_, S>) {
+            let span = ctx.span(id).expect("Missing span; this is a bug");
+            let mut lock = self.inner.lock().unwrap();
+            let is_removed = Arc::new(());
+            assert!(
+                lock.open
+                    .insert(span.name(), Arc::downgrade(&is_removed))
+                    .is_none(),
+                "test layer saw multiple spans with the same name, the test is probably messed up"
+            );
+            let mut extensions = span.extensions_mut();
+            extensions.insert(SetRemoved(is_removed));
+        }
+
+        fn on_close(&self, id: Id, ctx: Context<'_, S>) {
+            let span = if let Some(span) = ctx.span(&id) {
+                span
+            } else {
+                println!(
+                    "span {:?} did not exist in `on_close`, are we panicking?",
+                    id
+                );
+                return;
+            };
+            let name = span.name();
+            println!("close {} ({:?})", name, id);
+            if let Ok(mut lock) = self.inner.lock() {
+                if let Some(is_removed) = lock.open.remove(name) {
+                    assert!(is_removed.upgrade().is_some());
+                    lock.closed.push((name, is_removed));
+                }
+            }
+        }
+    }
+
+    impl CloseLayer {
+        fn new() -> (Self, CloseHandle) {
+            let state = Arc::new(Mutex::new(CloseState::default()));
+            (
+                Self {
+                    inner: state.clone(),
+                },
+                CloseHandle { state },
+            )
+        }
+    }
+
+    impl CloseState {
+        fn is_open(&self, span: &str) -> bool {
+            self.open.contains_key(span)
+        }
+
+        fn is_closed(&self, span: &str) -> bool {
+            self.closed.iter().any(|(name, _)| name == &span)
+        }
+    }
+
+    impl CloseHandle {
+        fn assert_closed(&self, span: &str) {
+            let lock = self.state.lock().unwrap();
+            assert!(
+                lock.is_closed(span),
+                "expected {} to be closed{}",
+                span,
+                if lock.is_open(span) {
+                    " (it was still open)"
+                } else {
+                    ", but it never existed (is there a problem with the test?)"
+                }
+            )
+        }
+
+        fn assert_open(&self, span: &str) {
+            let lock = self.state.lock().unwrap();
+            assert!(
+                lock.is_open(span),
+                "expected {} to be open{}",
+                span,
+                if lock.is_closed(span) {
+                    " (it was still open)"
+                } else {
+                    ", but it never existed (is there a problem with the test?)"
+                }
+            )
+        }
+
+        fn assert_removed(&self, span: &str) {
+            let lock = self.state.lock().unwrap();
+            let is_removed = match lock.closed.iter().find(|(name, _)| name == &span) {
+                Some((_, is_removed)) => is_removed,
+                None => panic!(
+                    "expected {} to be removed from the registry, but it was not closed {}",
+                    span,
+                    if lock.is_closed(span) {
+                        " (it was still open)"
+                    } else {
+                        ", but it never existed (is there a problem with the test?)"
+                    }
+                ),
+            };
+            assert!(
+                is_removed.upgrade().is_none(),
+                "expected {} to have been removed from the registry",
+                span
+            )
+        }
+
+        fn assert_not_removed(&self, span: &str) {
+            let lock = self.state.lock().unwrap();
+            let is_removed = match lock.closed.iter().find(|(name, _)| name == &span) {
+                Some((_, is_removed)) => is_removed,
+                None if lock.is_open(span) => return,
+                None => unreachable!(),
+            };
+            assert!(
+                is_removed.upgrade().is_some(),
+                "expected {} to have been removed from the registry",
+                span
+            )
+        }
+
+        #[allow(unused)] // may want this for future tests
+        fn assert_last_closed(&self, span: Option<&str>) {
+            let lock = self.state.lock().unwrap();
+            let last = lock.closed.last().map(|(span, _)| span);
+            assert_eq!(
+                last,
+                span.as_ref(),
+                "expected {:?} to have closed last",
+                span
+            );
+        }
+
+        fn assert_closed_in_order(&self, order: impl AsRef<[&'static str]>) {
+            let lock = self.state.lock().unwrap();
+            let order = order.as_ref();
+            for (i, name) in order.iter().enumerate() {
+                assert_eq!(
+                    lock.closed.get(i).map(|(span, _)| span),
+                    Some(name),
+                    "expected close order: {:?}, actual: {:?}",
+                    order,
+                    lock.closed.iter().map(|(name, _)| name).collect::<Vec<_>>()
+                );
+            }
+        }
+    }
+
+    #[test]
+    fn spans_are_removed_from_registry() {
+        let (close_layer, state) = CloseLayer::new();
+        let subscriber = AssertionLayer
+            .and_then(close_layer)
+            .with_subscriber(Registry::default());
+
+        // Create a `Dispatch` (which is internally reference counted) so that
+        // the subscriber lives to the end of the test. Otherwise, if we just
+        // passed the subscriber itself to `with_default`, we could see the span
+        // be dropped when the subscriber itself is dropped, destroying the
+        // registry.
+        let dispatch = dispatcher::Dispatch::new(subscriber);
+
+        dispatcher::with_default(&dispatch, || {
+            let span = tracing::debug_span!("span1");
+            drop(span);
+            let span = tracing::info_span!("span2");
+            drop(span);
+        });
+
+        state.assert_removed("span1");
+        state.assert_removed("span2");
+
+        // Ensure the registry itself outlives the span.
+        drop(dispatch);
+    }
+
+    #[test]
+    fn spans_are_only_closed_when_the_last_ref_drops() {
+        let (close_layer, state) = CloseLayer::new();
+        let subscriber = AssertionLayer
+            .and_then(close_layer)
+            .with_subscriber(Registry::default());
+
+        // Create a `Dispatch` (which is internally reference counted) so that
+        // the subscriber lives to the end of the test. Otherwise, if we just
+        // passed the subscriber itself to `with_default`, we could see the span
+        // be dropped when the subscriber itself is dropped, destroying the
+        // registry.
+        let dispatch = dispatcher::Dispatch::new(subscriber);
+
+        let span2 = dispatcher::with_default(&dispatch, || {
+            let span = tracing::debug_span!("span1");
+            drop(span);
+            let span2 = tracing::info_span!("span2");
+            let span2_clone = span2.clone();
+            drop(span2);
+            span2_clone
+        });
+
+        state.assert_removed("span1");
+        state.assert_not_removed("span2");
+
+        drop(span2);
+        state.assert_removed("span1");
+
+        // Ensure the registry itself outlives the span.
+        drop(dispatch);
+    }
+
+    #[test]
+    fn span_enter_guards_are_dropped_out_of_order() {
+        let (close_layer, state) = CloseLayer::new();
+        let subscriber = AssertionLayer
+            .and_then(close_layer)
+            .with_subscriber(Registry::default());
+
+        // Create a `Dispatch` (which is internally reference counted) so that
+        // the subscriber lives to the end of the test. Otherwise, if we just
+        // passed the subscriber itself to `with_default`, we could see the span
+        // be dropped when the subscriber itself is dropped, destroying the
+        // registry.
+        let dispatch = dispatcher::Dispatch::new(subscriber);
+
+        dispatcher::with_default(&dispatch, || {
+            let span1 = tracing::debug_span!("span1");
+            let span2 = tracing::info_span!("span2");
+
+            let enter1 = span1.enter();
+            let enter2 = span2.enter();
+
+            drop(enter1);
+            drop(span1);
+
+            state.assert_removed("span1");
+            state.assert_not_removed("span2");
+
+            drop(enter2);
+            state.assert_not_removed("span2");
+
+            drop(span2);
+            state.assert_removed("span1");
+            state.assert_removed("span2");
+        });
+    }
+
+    #[test]
+    fn child_closes_parent() {
+        // This test asserts that if a parent span's handle is dropped before
+        // a child span's handle, the parent will remain open until child
+        // closes, and will then be closed.
+
+        let (close_layer, state) = CloseLayer::new();
+        let subscriber = close_layer.with_subscriber(Registry::default());
+
+        let dispatch = dispatcher::Dispatch::new(subscriber);
+
+        dispatcher::with_default(&dispatch, || {
+            let span1 = tracing::info_span!("parent");
+            let span2 = tracing::info_span!(parent: &span1, "child");
+
+            state.assert_open("parent");
+            state.assert_open("child");
+
+            drop(span1);
+            state.assert_open("parent");
+            state.assert_open("child");
+
+            drop(span2);
+            state.assert_closed("parent");
+            state.assert_closed("child");
+        });
+    }
+
+    #[test]
+    fn child_closes_grandparent() {
+        // This test asserts that, when a span is kept open by a child which
+        // is *itself* kept open by a child, closing the grandchild will close
+        // both the parent *and* the grandparent.
+        let (close_layer, state) = CloseLayer::new();
+        let subscriber = close_layer.with_subscriber(Registry::default());
+
+        let dispatch = dispatcher::Dispatch::new(subscriber);
+
+        dispatcher::with_default(&dispatch, || {
+            let span1 = tracing::info_span!("grandparent");
+            let span2 = tracing::info_span!(parent: &span1, "parent");
+            let span3 = tracing::info_span!(parent: &span2, "child");
+
+            state.assert_open("grandparent");
+            state.assert_open("parent");
+            state.assert_open("child");
+
+            drop(span1);
+            drop(span2);
+            state.assert_open("grandparent");
+            state.assert_open("parent");
+            state.assert_open("child");
+
+            drop(span3);
+
+            state.assert_closed_in_order(["child", "parent", "grandparent"]);
+        });
+    }
+}

vendor/tracing-subscriber/src/registry/stack.rs

@@ -0,0 +1,77 @@
+pub(crate) use tracing_core::span::Id;
+
+#[derive(Debug)]
+struct ContextId {
+    id: Id,
+    duplicate: bool,
+}
+
+/// `SpanStack` tracks what spans are currently executing on a thread-local basis.
+///
+/// A "separate current span" for each thread is a semantic choice, as each span
+/// can be executing in a different thread.
+#[derive(Debug, Default)]
+pub(crate) struct SpanStack {
+    stack: Vec<ContextId>,
+}
+
+impl SpanStack {
+    #[inline]
+    pub(super) fn push(&mut self, id: Id) -> bool {
+        let duplicate = self.stack.iter().any(|i| i.id == id);
+        self.stack.push(ContextId { id, duplicate });
+        !duplicate
+    }
+
+    #[inline]
+    pub(super) fn pop(&mut self, expected_id: &Id) -> bool {
+        if let Some((idx, _)) = self
+            .stack
+            .iter()
+            .enumerate()
+            .rev()
+            .find(|(_, ctx_id)| ctx_id.id == *expected_id)
+        {
+            let ContextId { id: _, duplicate } = self.stack.remove(idx);
+            return !duplicate;
+        }
+        false
+    }
+
+    #[inline]
+    pub(crate) fn iter(&self) -> impl Iterator<Item = &Id> {
+        self.stack
+            .iter()
+            .rev()
+            .filter_map(|ContextId { id, duplicate }| if !*duplicate { Some(id) } else { None })
+    }
+
+    #[inline]
+    pub(crate) fn current(&self) -> Option<&Id> {
+        self.iter().next()
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::{Id, SpanStack};
+
+    #[test]
+    fn pop_last_span() {
+        let mut stack = SpanStack::default();
+        let id = Id::from_u64(1);
+        stack.push(id.clone());
+
+        assert!(stack.pop(&id));
+    }
+
+    #[test]
+    fn pop_first_span() {
+        let mut stack = SpanStack::default();
+        stack.push(Id::from_u64(1));
+        stack.push(Id::from_u64(2));
+
+        let id = Id::from_u64(1);
+        assert!(stack.pop(&id));
+    }
+}

vendor/tracing-subscriber/src/reload.rs

@@ -0,0 +1,394 @@
+//! Wrapper for a `Layer` to allow it to be dynamically reloaded.
+//!
+//! This module provides a [`Layer` type] implementing the [`Layer` trait] or [`Filter` trait]
+//! which wraps another type implementing the corresponding trait. This
+//! allows the wrapped type to be replaced with another
+//! instance of that type at runtime.
+//!
+//! This can be used in cases where a subset of `Layer` or `Filter` functionality
+//! should be dynamically reconfigured, such as when filtering directives may
+//! change at runtime. Note that this layer introduces a (relatively small)
+//! amount of overhead, and should thus only be used as needed.
+//!
+//! # Examples
+//!
+//! Reloading a [global filtering](crate::layer#global-filtering) layer:
+//!
+//! ```rust
+//! # use tracing::info;
+//! use tracing_subscriber::{filter, fmt, reload, prelude::*};
+//! let filter = filter::LevelFilter::WARN;
+//! let (filter, reload_handle) = reload::Layer::new(filter);
+//! tracing_subscriber::registry()
+//!   .with(filter)
+//!   .with(fmt::Layer::default())
+//!   .init();
+//! #
+//! # // specifying the Registry type is required
+//! # let _: &reload::Handle<filter::LevelFilter, tracing_subscriber::Registry> = &reload_handle;
+//! #
+//! info!("This will be ignored");
+//! reload_handle.modify(|filter| *filter = filter::LevelFilter::INFO);
+//! info!("This will be logged");
+//! ```
+//!
+//! Reloading a [`Filtered`](crate::filter::Filtered) layer:
+//!
+//! ```rust
+//! # use tracing::info;
+//! use tracing_subscriber::{filter, fmt, reload, prelude::*};
+//! let filtered_layer = fmt::Layer::default().with_filter(filter::LevelFilter::WARN);
+//! let (filtered_layer, reload_handle) = reload::Layer::new(filtered_layer);
+//! #
+//! # // specifying the Registry type is required
+//! # let _: &reload::Handle<filter::Filtered<fmt::Layer<tracing_subscriber::Registry>,
+//! # filter::LevelFilter, tracing_subscriber::Registry>,tracing_subscriber::Registry>
+//! # = &reload_handle;
+//! #
+//! tracing_subscriber::registry()
+//!   .with(filtered_layer)
+//!   .init();
+//! info!("This will be ignored");
+//! reload_handle.modify(|layer| *layer.filter_mut() = filter::LevelFilter::INFO);
+//! info!("This will be logged");
+//! ```
+//!
+//! ## Note
+//!
+//! The [`Layer`] implementation is unable to implement downcasting functionality,
+//! so certain [`Layer`] will fail to downcast if wrapped in a `reload::Layer`.
+//!
+//! If you only want to be able to dynamically change the
+//! `Filter` on a layer, prefer wrapping that `Filter` in the `reload::Layer`.
+//!
+//! [`Filter` trait]: crate::layer::Filter
+//! [`Layer` type]: Layer
+//! [`Layer` trait]: super::layer::Layer
+use crate::layer;
+use crate::sync::RwLock;
+
+use core::any::TypeId;
+use std::{
+    error, fmt,
+    marker::PhantomData,
+    sync::{Arc, Weak},
+};
+use tracing_core::{
+    callsite, span,
+    subscriber::{Interest, Subscriber},
+    Dispatch, Event, LevelFilter, Metadata,
+};
+
+/// Wraps a `Layer` or `Filter`, allowing it to be reloaded dynamically at runtime.
+#[derive(Debug)]
+pub struct Layer<L, S> {
+    // TODO(eliza): this once used a `crossbeam_util::ShardedRwLock`. We may
+    // eventually wish to replace it with a sharded lock implementation on top
+    // of our internal `RwLock` wrapper type. If possible, we should profile
+    // this first to determine if it's necessary.
+    inner: Arc<RwLock<L>>,
+    _s: PhantomData<fn(S)>,
+}
+
+/// Allows reloading the state of an associated [`Layer`](crate::layer::Layer).
+#[derive(Debug)]
+pub struct Handle<L, S> {
+    inner: Weak<RwLock<L>>,
+    _s: PhantomData<fn(S)>,
+}
+
+/// Indicates that an error occurred when reloading a layer.
+#[derive(Debug)]
+pub struct Error {
+    kind: ErrorKind,
+}
+
+#[derive(Debug)]
+enum ErrorKind {
+    SubscriberGone,
+    Poisoned,
+}
+
+// ===== impl Layer =====
+
+impl<L, S> crate::Layer<S> for Layer<L, S>
+where
+    L: crate::Layer<S> + 'static,
+    S: Subscriber,
+{
+    fn on_register_dispatch(&self, subscriber: &Dispatch) {
+        try_lock!(self.inner.read()).on_register_dispatch(subscriber);
+    }
+
+    fn on_layer(&mut self, subscriber: &mut S) {
+        try_lock!(self.inner.write(), else return).on_layer(subscriber);
+    }
+
+    #[inline]
+    fn register_callsite(&self, metadata: &'static Metadata<'static>) -> Interest {
+        try_lock!(self.inner.read(), else return Interest::sometimes()).register_callsite(metadata)
+    }
+
+    #[inline]
+    fn enabled(&self, metadata: &Metadata<'_>, ctx: layer::Context<'_, S>) -> bool {
+        try_lock!(self.inner.read(), else return false).enabled(metadata, ctx)
+    }
+
+    #[inline]
+    fn on_new_span(&self, attrs: &span::Attributes<'_>, id: &span::Id, ctx: layer::Context<'_, S>) {
+        try_lock!(self.inner.read()).on_new_span(attrs, id, ctx)
+    }
+
+    #[inline]
+    fn on_record(&self, span: &span::Id, values: &span::Record<'_>, ctx: layer::Context<'_, S>) {
+        try_lock!(self.inner.read()).on_record(span, values, ctx)
+    }
+
+    #[inline]
+    fn on_follows_from(&self, span: &span::Id, follows: &span::Id, ctx: layer::Context<'_, S>) {
+        try_lock!(self.inner.read()).on_follows_from(span, follows, ctx)
+    }
+
+    #[inline]
+    fn event_enabled(&self, event: &Event<'_>, ctx: layer::Context<'_, S>) -> bool {
+        try_lock!(self.inner.read(), else return false).event_enabled(event, ctx)
+    }
+
+    #[inline]
+    fn on_event(&self, event: &Event<'_>, ctx: layer::Context<'_, S>) {
+        try_lock!(self.inner.read()).on_event(event, ctx)
+    }
+
+    #[inline]
+    fn on_enter(&self, id: &span::Id, ctx: layer::Context<'_, S>) {
+        try_lock!(self.inner.read()).on_enter(id, ctx)
+    }
+
+    #[inline]
+    fn on_exit(&self, id: &span::Id, ctx: layer::Context<'_, S>) {
+        try_lock!(self.inner.read()).on_exit(id, ctx)
+    }
+
+    #[inline]
+    fn on_close(&self, id: span::Id, ctx: layer::Context<'_, S>) {
+        try_lock!(self.inner.read()).on_close(id, ctx)
+    }
+
+    #[inline]
+    fn on_id_change(&self, old: &span::Id, new: &span::Id, ctx: layer::Context<'_, S>) {
+        try_lock!(self.inner.read()).on_id_change(old, new, ctx)
+    }
+
+    #[inline]
+    fn max_level_hint(&self) -> Option<LevelFilter> {
+        try_lock!(self.inner.read(), else return None).max_level_hint()
+    }
+
+    #[doc(hidden)]
+    unsafe fn downcast_raw(&self, id: TypeId) -> Option<*const ()> {
+        // Safety: it is generally unsafe to downcast through a reload, because
+        // the pointer can be invalidated after the lock is dropped.
+        // `NoneLayerMarker` is a special case because it
+        // is never dereferenced.
+        //
+        // Additionally, even if the marker type *is* dereferenced (which it
+        // never will be), the pointer should be valid even if the subscriber
+        // is reloaded, because all `NoneLayerMarker` pointers that we return
+        // actually point to the global static singleton `NoneLayerMarker`,
+        // rather than to a field inside the lock.
+        if id == TypeId::of::<layer::NoneLayerMarker>() {
+            return try_lock!(self.inner.read(), else return None).downcast_raw(id);
+        }
+
+        None
+    }
+}
+
+// ===== impl Filter =====
+
+#[cfg(all(feature = "registry", feature = "std"))]
+#[cfg_attr(docsrs, doc(cfg(all(feature = "registry", feature = "std"))))]
+impl<S, L> crate::layer::Filter<S> for Layer<L, S>
+where
+    L: crate::layer::Filter<S> + 'static,
+    S: Subscriber,
+{
+    #[inline]
+    fn callsite_enabled(&self, metadata: &'static Metadata<'static>) -> Interest {
+        try_lock!(self.inner.read(), else return Interest::sometimes()).callsite_enabled(metadata)
+    }
+
+    #[inline]
+    fn enabled(&self, metadata: &Metadata<'_>, ctx: &layer::Context<'_, S>) -> bool {
+        try_lock!(self.inner.read(), else return false).enabled(metadata, ctx)
+    }
+
+    #[inline]
+    fn on_new_span(&self, attrs: &span::Attributes<'_>, id: &span::Id, ctx: layer::Context<'_, S>) {
+        try_lock!(self.inner.read()).on_new_span(attrs, id, ctx)
+    }
+
+    #[inline]
+    fn on_record(&self, span: &span::Id, values: &span::Record<'_>, ctx: layer::Context<'_, S>) {
+        try_lock!(self.inner.read()).on_record(span, values, ctx)
+    }
+
+    #[inline]
+    fn on_enter(&self, id: &span::Id, ctx: layer::Context<'_, S>) {
+        try_lock!(self.inner.read()).on_enter(id, ctx)
+    }
+
+    #[inline]
+    fn on_exit(&self, id: &span::Id, ctx: layer::Context<'_, S>) {
+        try_lock!(self.inner.read()).on_exit(id, ctx)
+    }
+
+    #[inline]
+    fn on_close(&self, id: span::Id, ctx: layer::Context<'_, S>) {
+        try_lock!(self.inner.read()).on_close(id, ctx)
+    }
+
+    #[inline]
+    fn max_level_hint(&self) -> Option<LevelFilter> {
+        try_lock!(self.inner.read(), else return None).max_level_hint()
+    }
+}
+
+impl<L, S> Layer<L, S> {
+    /// Wraps the given [`Layer`] or [`Filter`], returning a `reload::Layer`
+    /// and a `Handle` that allows the inner value to be modified at runtime.
+    ///
+    /// [`Layer`]: crate::layer::Layer
+    /// [`Filter`]: crate::layer::Filter
+    pub fn new(inner: L) -> (Self, Handle<L, S>) {
+        let this = Self {
+            inner: Arc::new(RwLock::new(inner)),
+            _s: PhantomData,
+        };
+        let handle = this.handle();
+        (this, handle)
+    }
+
+    /// Returns a `Handle` that can be used to reload the wrapped [`Layer`] or [`Filter`].
+    ///
+    /// [`Layer`]: crate::layer::Layer
+    /// [`Filter`]: crate::layer::Filter
+    pub fn handle(&self) -> Handle<L, S> {
+        Handle {
+            inner: Arc::downgrade(&self.inner),
+            _s: PhantomData,
+        }
+    }
+}
+
+// ===== impl Handle =====
+
+impl<L, S> Handle<L, S> {
+    /// Replace the current [`Layer`] or [`Filter`] with the provided `new_value`.
+    ///
+    /// [`Handle::reload`] cannot be used with the [`Filtered`] layer; use
+    /// [`Handle::modify`] instead (see [this issue] for additional details).
+    ///
+    /// However, if the _only_ the [`Filter`]  needs to be modified, use
+    /// `reload::Layer` to wrap the `Filter` directly.
+    ///
+    /// [`Layer`]: crate::layer::Layer
+    /// [`Filter`]: crate::layer::Filter
+    /// [`Filtered`]: crate::filter::Filtered
+    ///
+    /// [this issue]: https://github.com/tokio-rs/tracing/issues/1629
+    pub fn reload(&self, new_value: impl Into<L>) -> Result<(), Error> {
+        self.modify(|layer| {
+            *layer = new_value.into();
+        })
+    }
+
+    /// Invokes a closure with a mutable reference to the current layer or filter,
+    /// allowing it to be modified in place.
+    pub fn modify(&self, f: impl FnOnce(&mut L)) -> Result<(), Error> {
+        let inner = self.inner.upgrade().ok_or(Error {
+            kind: ErrorKind::SubscriberGone,
+        })?;
+
+        let mut lock = try_lock!(inner.write(), else return Err(Error::poisoned()));
+        f(&mut *lock);
+        // Release the lock before rebuilding the interest cache, as that
+        // function will lock the new layer.
+        drop(lock);
+
+        callsite::rebuild_interest_cache();
+
+        // If the `log` crate compatibility feature is in use, set `log`'s max
+        // level as well, in case the max `tracing` level changed. We do this
+        // *after* rebuilding the interest cache, as that's when the `tracing`
+        // max level filter is re-computed.
+        #[cfg(feature = "tracing-log")]
+        tracing_log::log::set_max_level(tracing_log::AsLog::as_log(
+            &crate::filter::LevelFilter::current(),
+        ));
+
+        Ok(())
+    }
+
+    /// Returns a clone of the layer or filter's current value if it still exists.
+    /// Otherwise, if the subscriber has been dropped, returns `None`.
+    pub fn clone_current(&self) -> Option<L>
+    where
+        L: Clone,
+    {
+        self.with_current(L::clone).ok()
+    }
+
+    /// Invokes a closure with a borrowed reference to the current layer or filter,
+    /// returning the result (or an error if the subscriber no longer exists).
+    pub fn with_current<T>(&self, f: impl FnOnce(&L) -> T) -> Result<T, Error> {
+        let inner = self.inner.upgrade().ok_or(Error {
+            kind: ErrorKind::SubscriberGone,
+        })?;
+        let inner = try_lock!(inner.read(), else return Err(Error::poisoned()));
+        Ok(f(&*inner))
+    }
+}
+
+impl<L, S> Clone for Handle<L, S> {
+    fn clone(&self) -> Self {
+        Handle {
+            inner: self.inner.clone(),
+            _s: PhantomData,
+        }
+    }
+}
+
+// ===== impl Error =====
+
+impl Error {
+    fn poisoned() -> Self {
+        Self {
+            kind: ErrorKind::Poisoned,
+        }
+    }
+
+    /// Returns `true` if this error occurred because the layer was poisoned by
+    /// a panic on another thread.
+    pub fn is_poisoned(&self) -> bool {
+        matches!(self.kind, ErrorKind::Poisoned)
+    }
+
+    /// Returns `true` if this error occurred because the `Subscriber`
+    /// containing the reloadable layer was dropped.
+    pub fn is_dropped(&self) -> bool {
+        matches!(self.kind, ErrorKind::SubscriberGone)
+    }
+}
+
+impl fmt::Display for Error {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        let msg = match self.kind {
+            ErrorKind::SubscriberGone => "subscriber no longer exists",
+            ErrorKind::Poisoned => "lock poisoned",
+        };
+        f.pad(msg)
+    }
+}
+
+impl error::Error for Error {}

vendor/tracing-subscriber/src/sync.rs

@@ -0,0 +1,65 @@
+//! Abstracts over sync primitive implementations.
+//!
+//! Optionally, we allow the Rust standard library's `RwLock` to be replaced
+//! with the `parking_lot` crate's implementation. This may provide improved
+//! performance in some cases. However, the `parking_lot` dependency is an
+//! opt-in feature flag. Because `parking_lot::RwLock` has a slightly different
+//! API than `std::sync::RwLock` (it does not support poisoning on panics), we
+//! wrap it with a type that provides the same method signatures. This allows us
+//! to transparently swap `parking_lot` in without changing code at the callsite.
+#[allow(unused_imports)] // may be used later;
+pub(crate) use std::sync::{LockResult, PoisonError, TryLockResult};
+
+#[cfg(not(feature = "parking_lot"))]
+pub(crate) use std::sync::{RwLock, RwLockReadGuard, RwLockWriteGuard};
+
+#[cfg(feature = "parking_lot")]
+pub(crate) use self::parking_lot_impl::*;
+
+#[cfg(feature = "parking_lot")]
+mod parking_lot_impl {
+    pub(crate) use parking_lot::{RwLockReadGuard, RwLockWriteGuard};
+    use std::sync::{LockResult, TryLockError, TryLockResult};
+
+    #[derive(Debug)]
+    pub(crate) struct RwLock<T> {
+        inner: parking_lot::RwLock<T>,
+    }
+
+    impl<T> RwLock<T> {
+        pub(crate) fn new(val: T) -> Self {
+            Self {
+                inner: parking_lot::RwLock::new(val),
+            }
+        }
+
+        #[inline]
+        pub(crate) fn get_mut(&mut self) -> LockResult<&mut T> {
+            Ok(self.inner.get_mut())
+        }
+
+        #[inline]
+        pub(crate) fn read(&self) -> LockResult<RwLockReadGuard<'_, T>> {
+            Ok(self.inner.read())
+        }
+
+        #[inline]
+        #[allow(dead_code)] // may be used later;
+        pub(crate) fn try_read(&self) -> TryLockResult<RwLockReadGuard<'_, T>> {
+            self.inner.try_read().ok_or(TryLockError::WouldBlock)
+        }
+
+        #[inline]
+        pub(crate) fn write(&self) -> LockResult<RwLockWriteGuard<'_, T>> {
+            Ok(self.inner.write())
+        }
+    }
+
+    impl<T: Default> Default for RwLock<T> {
+        fn default() -> Self {
+            RwLock {
+                inner: parking_lot::RwLock::default(),
+            }
+        }
+    }
+}

vendor/tracing-subscriber/src/util.rs

@@ -0,0 +1,153 @@
+//! Extension traits and other utilities to make working with subscribers more
+//! ergonomic.
+use core::fmt;
+#[cfg(feature = "std")]
+use std::error::Error;
+use tracing_core::dispatcher::{self, Dispatch};
+#[cfg(feature = "tracing-log")]
+use tracing_log::AsLog;
+
+/// Extension trait adding utility methods for subscriber initialization.
+///
+/// This trait provides extension methods to make configuring and setting a
+/// [default subscriber] more ergonomic. It is automatically implemented for all
+/// types that can be converted into a [trace dispatcher]. Since `Dispatch`
+/// implements `From<T>` for all `T: Subscriber`, all `Subscriber`
+/// implementations will implement this extension trait as well. Types which
+/// can be converted into `Subscriber`s, such as builders that construct a
+/// `Subscriber`, may implement `Into<Dispatch>`, and will also receive an
+/// implementation of this trait.
+///
+/// [default subscriber]: https://docs.rs/tracing/0.1.21/tracing/dispatcher/index.html#setting-the-default-subscriber
+/// [trace dispatcher]: https://docs.rs/tracing/0.1.21/tracing/dispatcher/index.html
+pub trait SubscriberInitExt
+where
+    Self: Into<Dispatch>,
+{
+    /// Sets `self` as the [default subscriber] in the current scope, returning a
+    /// guard that will unset it when dropped.
+    ///
+    /// If the "tracing-log" feature flag is enabled, this will also initialize
+    /// a [`log`] compatibility layer. This allows the subscriber to consume
+    /// `log::Record`s as though they were `tracing` `Event`s.
+    ///
+    /// [default subscriber]: https://docs.rs/tracing/0.1.21/tracing/dispatcher/index.html#setting-the-default-subscriber
+    /// [`log`]: https://crates.io/log
+    #[cfg(feature = "std")]
+    #[cfg_attr(docsrs, doc(cfg(feature = "std")))]
+    fn set_default(self) -> dispatcher::DefaultGuard {
+        #[cfg(feature = "tracing-log")]
+        let _ = tracing_log::LogTracer::init();
+
+        dispatcher::set_default(&self.into())
+    }
+
+    /// Attempts to set `self` as the [global default subscriber] in the current
+    /// scope, returning an error if one is already set.
+    ///
+    /// If the "tracing-log" feature flag is enabled, this will also attempt to
+    /// initialize a [`log`] compatibility layer. This allows the subscriber to
+    /// consume `log::Record`s as though they were `tracing` `Event`s.
+    ///
+    /// This method returns an error if a global default subscriber has already
+    /// been set, or if a `log` logger has already been set (when the
+    /// "tracing-log" feature is enabled).
+    ///
+    /// [global default subscriber]: https://docs.rs/tracing/0.1.21/tracing/dispatcher/index.html#setting-the-default-subscriber
+    /// [`log`]: https://crates.io/log
+    fn try_init(self) -> Result<(), TryInitError> {
+        dispatcher::set_global_default(self.into()).map_err(TryInitError::new)?;
+
+        // Since we are setting the global default subscriber, we can
+        // opportunistically go ahead and set its global max level hint as
+        // the max level for the `log` crate as well. This should make
+        // skipping `log` diagnostics much faster.
+        #[cfg(feature = "tracing-log")]
+        tracing_log::LogTracer::builder()
+            // Note that we must call this *after* setting the global default
+            // subscriber, so that we get its max level hint.
+            .with_max_level(tracing_core::LevelFilter::current().as_log())
+            .init()
+            .map_err(TryInitError::new)?;
+
+        Ok(())
+    }
+
+    /// Attempts to set `self` as the [global default subscriber] in the current
+    /// scope, panicking if this fails.
+    ///
+    /// If the "tracing-log" feature flag is enabled, this will also attempt to
+    /// initialize a [`log`] compatibility layer. This allows the subscriber to
+    /// consume `log::Record`s as though they were `tracing` `Event`s.
+    ///
+    /// This method panics if a global default subscriber has already been set,
+    /// or if a `log` logger has already been set (when the "tracing-log"
+    /// feature is enabled).
+    ///
+    /// [global default subscriber]: https://docs.rs/tracing/0.1.21/tracing/dispatcher/index.html#setting-the-default-subscriber
+    /// [`log`]: https://crates.io/log
+    fn init(self) {
+        self.try_init()
+            .expect("failed to set global default subscriber")
+    }
+}
+
+impl<T> SubscriberInitExt for T where T: Into<Dispatch> {}
+
+/// Error returned by [`try_init`](SubscriberInitExt::try_init) if a global default subscriber could not be initialized.
+pub struct TryInitError {
+    #[cfg(feature = "std")]
+    inner: Box<dyn Error + Send + Sync + 'static>,
+
+    #[cfg(not(feature = "std"))]
+    _p: (),
+}
+
+// ==== impl TryInitError ====
+
+impl TryInitError {
+    #[cfg(feature = "std")]
+    fn new(e: impl Into<Box<dyn Error + Send + Sync + 'static>>) -> Self {
+        Self { inner: e.into() }
+    }
+
+    #[cfg(not(feature = "std"))]
+    fn new<T>(_: T) -> Self {
+        Self { _p: () }
+    }
+}
+
+impl fmt::Debug for TryInitError {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        #[cfg(feature = "std")]
+        {
+            fmt::Debug::fmt(&self.inner, f)
+        }
+
+        #[cfg(not(feature = "std"))]
+        {
+            f.write_str("TryInitError(())")
+        }
+    }
+}
+
+impl fmt::Display for TryInitError {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        #[cfg(feature = "std")]
+        {
+            fmt::Display::fmt(&self.inner, f)
+        }
+
+        #[cfg(not(feature = "std"))]
+        {
+            f.write_str("failed to set global default subscriber")
+        }
+    }
+}
+
+#[cfg(feature = "std")]
+impl Error for TryInitError {
+    fn source(&self) -> Option<&(dyn Error + 'static)> {
+        self.inner.source()
+    }
+}