chore(deps): update patch crates

[renovate]

This PR contains the following updates:

Package	Type	Update	Change
arc-swap	dependencies	patch	1.8.2 → 1.9.1
cfg-if	dependencies	patch	1.0 → 1.0.4
document-features (source)	dependencies	patch	0.2.8 → 0.2.12
futures (source)	dependencies	patch	0.3.31 → 0.3.32
indexmap	dependencies	patch	2.12.0 → 2.14.0
json-strip-comments	dependencies	patch	3.0.1 → 3.1.1
libfuzzer-sys	dependencies	patch	0.4.7 → 0.4.12
pnp (source)	dependencies	patch	0.12.5 → 0.12.9
rayon	dev-dependencies	patch	1.11.0 → 1.12.0
regex	dependencies	patch	1.11.1 → 1.12.3
regex	dev-dependencies	patch	1.12.2 → 1.12.3
rustc-hash	dependencies	patch	2.1.1 → 2.1.2
serde_json	dependencies	patch	1.0.145 → 1.0.150
thiserror	dependencies	patch	2.0.17 → 2.0.18
tokio (source)	dependencies	patch	1.48.0 → 1.52.3
tracing (source)	dependencies	patch	0.1.41 → 0.1.44
tracing-subscriber (source)	dependencies	patch	0.3.18 → 0.3.23

---

Release Notes

rust-fuzz/libfuzzer (libfuzzer-sys)

v0.4.12

Released 2026-02-10.

Changed

• Recommend SmallRng over StdRng in the examples for faster, more lightweight
  seeding and sampling
• Updated rand dependency from 0.8.5 to 0.10
• Updated flate2 dependency from 1.0.24 to 1.1
• Rename gen variable to rng for better 2024 Edition compatibility

---

v0.4.11

Compare Source

Released 2026-02-10.

Changed

• Updated to libFuzzer commit a47b42eb9f9b (release/22.x).

---

v0.4.10

Compare Source

Released 2025-07-03.

Changed

• Updated to libFuzzer commit 6146a88f6049 (release/20.x).
• Fuzz targets taking raw byte slice inputs can now return Corpus results.

---

v0.4.9

Compare Source

Released 2025-01-28.

Added

• The example_init demonstrates how to pass an initialization code block to
  the fuzz_target! macro.

Changed

• The fuzz_target! macro now supports the generation of LLVMFuzzerInitialize
  to execute initialization code once before running the fuzzer. This change is
  not breaking and is completely backward compatible.

---

v0.4.8

Compare Source

Released 2024-11-07.

Added

• Bindings to LLVMFuzzerCustomCrossOver through the fuzz_crossover
  macro. See the example_crossover directory in this crate's repo for a
  complete example.

Changed

• Updated to libFuzzer commit ab51eccf88f5321e7c60591c5546b254b6afab99
  (release/19.x).

---

yarnpkg/pnp-rs (pnp)

v0.12.9

Compare Source

Fixed

• (fs) preserve drive root in virtual paths (#​106)
• (util) keep drive prefix sticky when .. overshoots drive root (#​104)

Other

• clean up duplicated code (#​107)
• (deps) update rust crate fancy-regex to ^0.18.0 (#​103)
• (deps) update yarn to v4.14.1 (#​102)
• (deps) update yarn to v4.14.0 (#​101)
• (deps) update dependency rust to v1.95.0 (#​100)
• (deps) update rust crate indexmap to v2.14.0 (#​99)
• (deps) update rust crate indexmap to v2.13.1 (#​98)
• (deps) update rust crate rustc-hash to v2.1.2 (#​97)
• (deps) update dependency rust to v1.94.1 (#​96)
• (deps) update yarn to v4.13.0 (#​95)
• (deps) update dependency rust to v1.94.0 (#​94)
• (deps) update dependency rust to v1.93.1 (#​93)
• (deps) update rust crates (#​92)
• (deps) update dependency rust to v1.93.0 (#​91)

serde-rs/json (serde_json)

v1.0.150

Compare Source

• Reject non-string enum object keys (#​1324, thanks @​puneetdixit200)

---

Configuration

📅 Schedule: (in timezone Asia/Shanghai)

• Branch creation
  • "before 8am on saturday"
• Automerge
  • At any time (no schedule defined)

🚦 Automerge: Disabled by config. Please merge this manually once you are satisfied.

♻ Rebasing: Whenever PR becomes conflicted, or you tick the rebase/retry checkbox.

👻 Immortal: This PR will be recreated if closed unmerged. Get config help if that's undesired.

---

• If you want to rebase/retry this PR, check this box

---

This PR was generated by Mend Renovate. View the repository job log.

[codspeed-hq]

Merging this PR will improve performance by 4.2%

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

⚡ 2 improved benchmarks
✅ 8 untouched benchmarks

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
⚡	Simulation	resolver[[single-threaded]resolve with many extensions]	134.2 ms	127.8 ms	+5.01%
⚡	Simulation	resolver[pnp resolve]	258.2 µs	249.7 µs	+3.41%

Tip

Curious why this is faster? Comment @codspeedbot explain why this is faster on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing renovate/patch-crates (0b44f97) with main (78cb104)}

[stormslowly]

@codspeedbot explain why this is faster

[codspeed-hq]

@codspeedbot explain why this is faster

Why this is faster

The 4.8% improvement in resolver[[single-threaded]resolve with many extensions] is driven by the pnp crate upgrade from 0.12.8 → 0.12.9.

Root cause: optimized string splitting in pnp::fs::VPath::from

The VPath::from function is a hot path in this benchmark — it's called during every filesystem metadata lookup to classify virtual paths. Comparing the flamegraphs between BASE and HEAD reveals a clear change in how path components are split:

	BASE (pnp@0.12.8)	HEAD (pnp@0.12.9)
Split pattern type	&[char] (char slice)	[char; 2] (fixed-size array)
SplitInternal::next self time	11.3 ms (8.42%)	3.6 ms (2.84%)
SplitInternal::next total time	33.2 ms (24.80%)	12 ms (9.36%)
VPath::from total time	53.7 ms (40.08%)	46 ms (35.98%)

In the BASE version, splitting with &[char] uses Rust's MultiCharEqSearcher, which iterates through the char slice for every position — involving calls to any<char, ...>, matches, and next<&[char]> that together consumed ~9.1 ms of self time.

In the HEAD version, splitting with [char; 2] allows the compiler to emit a direct two-way comparison instead of a loop, eliminating that overhead entirely. The SplitInternal::next total time dropped by ~21 ms, saving roughly 6.1 ms at the benchmark level after accounting for redistribution of inlined work.

Note on environment

The runs were executed on different hardware (Intel Xeon vs AMD EPYC), but since CodSpeed uses simulation mode (instruction counting), the measurement is hardware-independent — this improvement reflects a real reduction in executed instructions, not a hardware artifact.

Summary

• Warming up the performance wizard...
• Analyzing performance improvement — Flamegraph comparison complete — root cause identified
• Handling comment — Comment handling finished successfully

_{Was this response helpful? React with 👍 or 👎 to help us improve CodSpeedBot.}