memvid

Bug fixes

#230 — WAL checksum mismatch after region growth

Sustained put + commit workloads that grew the embedded WAL could corrupt it, surfacing as Embedded WAL is corrupted at offset N: wal record checksum mismatch (and, in some reports, runaway sparse-file growth past EOF).

grow_wal_region / ensure_wal_capacity shifted the data region right by delta and updated data_end/footer_offset, but left cached_payload_end stale. rebuild_indexes then seeked to that pre-growth offset — now inside the grown WAL region — and overwrote live WAL record payloads. cached_payload_end is now advanced by the same delta, restoring the cached_payload_end >= wal_offset + wal_size invariant.

#215 — Tantivy working directories leaked to the system temp dir

Each put() that rebuilt the Tantivy index created a temporary working directory that was never cleaned up, accumulating indefinitely (notably on Windows). TantivyEngine declared its TempDir first, so it was dropped before the index/reader/writer released their open file handles and lock files — making the directory removal silently fail. The TempDir now drops last and the index writer is released first, so per-put working directories are cleaned up.

A commit_per_put_survives_wal_growth regression test covers the #230 path across multiple WAL growth cycles.

Full Changelog: https://github.com/memvid/memvid/compare/v2.0.139...v2.0.140