tracing-subscriber tests fail with --all-features and chrono 0.4.42

[plugwash]

Bug Report

Version

tracing-subscriber 0.3.20 and latest git

Platform

Linux 5.10 (debian sid chroot)

Crates

tracing-subscriber

Description

git clone https://github.com/tokio-rs/tracing
cd tracing/tracing-subscriber/
cargo test --all-features

results in

failures:

---- fmt::time::chrono_crate::tests::test_chrono_format_time_local_default stdout ----

thread 'fmt::time::chrono_crate::tests::test_chrono_format_time_local_default' panicked at tracing-subscriber/src/fmt/time/chrono_crate.rs:158:9:
assertion failed: chrono::DateTime::parse_from_str(&buf, "%FT%H:%M:%S%.6f%z").is_ok()
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

---- fmt::time::chrono_crate::tests::test_chrono_format_time_utc_default stdout ----

thread 'fmt::time::chrono_crate::tests::test_chrono_format_time_utc_default' panicked at tracing-subscriber/src/fmt/time/chrono_crate.rs:137:9:
assertion failed: chrono::DateTime::parse_from_str(&buf, "%FT%H:%M:%S%.6f%z").is_ok()


failures:
    fmt::time::chrono_crate::tests::test_chrono_format_time_local_default
    fmt::time::chrono_crate::tests::test_chrono_format_time_utc_default

Downgrading chrono to 0.4.41 (with "cargo update -p chrono --precise 0.4.41") results in the tests passing.

[hds]

Would you be able to provide the backtrace for this error?

Maybe we've always been using chrono wrong, but this does seem like a breaking change in that chrono release, maybe there's an issue for that?

[plugwash]

stack backtrace:
0: __rustc::rust_begin_unwind
at /usr/src/rustc-1.90.0/library/std/src/panicking.rs:697:5
1: core::panicking::panic_fmt
at /usr/src/rustc-1.90.0/library/core/src/panicking.rs:75:14
2: core::panicking::panic
at /usr/src/rustc-1.90.0/library/core/src/panicking.rs:145:5
3: tracing_subscriber::fmt::time::chrono_crate::tests::test_chrono_format_time_utc_default
at ./src/fmt/time/chrono_crate.rs:137:9
4: tracing_subscriber::fmt::time::chrono_crate::tests::test_chrono_format_time_utc_default::{{closure}}
at ./src/fmt/time/chrono_crate.rs:132:45
5: core::ops::function::FnOnce::call_once
at /usr/src/rustc-1.90.0/library/core/src/ops/function.rs:253:5
6: core::ops::function::FnOnce::call_once
at /usr/src/rustc-1.90.0/library/core/src/ops/function.rs:253:5

I added a print statement and the printed timestamp had the same format as described in comments, so this seems to be an incompatible change in the chrono::DateTime::parse_from_str function used by the test.

[hds]

I couldn't repro locally (but I'm on macOS). I also realised that these tests don't seem to be running on CI. I'm going to try and fix that and see if it can repro there.

However, maybe if you can reproduce outside tracing it would be worth submitting an issue to chrono?