pydantic/jiter - CodSpeed

jiter

Blog Docs Changelog

Performance History

Latest Results

fix(jiter): honour partial mode in `StringDecoderRange` escape handling (#272) Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>

eouzoe:pr-272-partial-bytes

2 days ago

fix(jiter): bound `current_index` in partial string mode (#267) Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>

eouzoe:pr-267-partial-index

2 days ago

Merge remote-tracking branch 'origin/main' into structural-skip # Conflicts: # crates/jiter/tests/main.rs

structural-skip

8 days ago

Add `Jiter::with_tape` and `Jiter::into_tape` (#298) Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>

main

8 days ago

Make the pool test prove the caches still work The round-trip assertion could not fail: every cache was out of the pool at that point, so its usage was zero whatever the pool had done. Each cache now interns a string, checks a second lookup returns the same object, and is checked to still hold it once back in the pool. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

string-cache-pool

8 days ago

Scan number digit runs in `NumberRange` with the SIMD digit-run scan (#300) Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>

main

8 days ago

Cover the pool's full and empty edges Nothing exercised the branch that drops a returned cache when every slot is taken, nor taking from an empty pool, which is what coverage flagged. The test takes more caches than there are slots, hands them all back, and checks the pool kept a slotful, dropped the rest, and is empty again once they are taken out. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

string-cache-pool

8 days ago

Empty one pool slot at a time, and serialise the cache tests `cache_usage` and `cache_clear` took every cache out of the pool before handing any back, so a parse starting meanwhile found an empty pool and built its own, and one finishing during the refill could have its cache dropped although the pool had room. Each cache now goes straight back into the slot it came from, so only one of the eight is ever out. The cache is process-global and the harness runs this binary's tests on several threads, so the tests that parse with it on now take a lock, and clear it while attached to the interpreter, since dropping cached strings releases Python objects. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

string-cache-pool

8 days ago

Latest Branches

-1%

fix(jiter): honour partial mode in `StringDecoderRange` escape handling (#272)#306

2 days ago

8f935a6

eouzoe:pr-272-partial-bytes

+10%

fix(jiter): bound `current_index` in partial string mode (#267)#305

2 days ago

6aeeeaf

eouzoe:pr-267-partial-index

+56%

Skip arrays and objects with a structural bitmap scan#301

8 days ago

711ec91

structural-skip

© 2026 CodSpeed Technology

Home Terms Privacy Docs