Hardening Allocations & Concurrency - #764

Merged

ashvardanian merged 18 commits into

mainfrom

May 24, 2026

Commits on May 23, 2026

Fix: Eager-reserve thread contexts in `index_dense_gt::make` (#755)

authored
b296566
Fix: Resize cast buffer in `change_metric` for new bytes-per-vector
ashvardanian
committed
d544745
Fix: Bounded probe in `equal_iterator_gt::operator++`
ashvardanian
committed
b779c47
Fix: Report OOM from C-ABI thread-limit changers
ashvardanian
committed
ad24056
Fix: Stop C-ABI metadata readers on failure
ashvardanian
committed
34889ee
Fix: Stop `usearch_init` on `make` failure
ashvardanian
committed
2aa0070
Fix: Stop JavaScript `Remove` loop after exception throw
ashvardanian
committed
f3e1052
Fix: Restore `ring_gt::try_push` return value
ashvardanian
committed
18c44ee
Improve: Release Python GIL during long index operations
ashvardanian
committed
d8be67d
Improve: Test GIL-release contract and progress-callback path
ashvardanian
committed
a598493
Improve: Serialize concurrent same-`Index` Python access with a mutex
ashvardanian
committed
47528b5

Commits on May 24, 2026

Fix: Keep `vectors_lookup_` capacity after `clear()` #759

authored
9d77be5
Fix: Short-circuit self-renames (#761)

authored
830f31a
Fix: Refuse missing metrics in C change-metric API (#760)

authored
490c1b2
Fix: Guard quantized casts against zero-magnitude inputs (#758)

authored
0c903f4
Fix: Preserve hash lookup capacity across thread reserves (#765)

authored
3cf843b
Fix: Checked arithmetic for allocation sizes (#763)

authored
89da7c5
Fix: Refuse operations without reserved thread contexts (#757)

authored
b8d3403