Skip to content
zit Timeline
Login

Timeline

10 most recent check-ins occurring on or after 2026-07-28 11:49:20.

More ↑
2026-07-29
14:19
merge logstore-index: checksum the store through a vendored crc32

std.hash.Crc32 is byte-at-a-time over one table (~350 MB/s here) and the open-time scan CRCs every payload byte, so it was ~93% of a cold open. Slicing-by-8 in its place: a 416 MB store opens in 292ms against 1.22s, warm reads run 3.3x, the write path 1.5x. Bit-identical by construction, so no format changes and no repository is affected. check-in: 602c2459f2 user: dmestas tags: trunk

13:21
crc32: export for the benchmark and the fuzz lane

Both consumers live in their own build roots and cannot reach crc32.zig by path, so the export is forced. Adds the differential fuzz lane against std — the only coverage that can see a self-consistent but numerically shifted implementation, which would orphan every existing .zit file.

The implementation brief prohibited this export, citing repocache as the precedent for unexported leaf modules. That precedent does not apply: repocache has no out-of-module consumer. storestat and repoimage, which do, are both exported. Overridden deliberately. closed check-in: c57c16c5cd user: dmestas tags: logstore-index

13:19
repocache, bisect, undo: checksum through the vendored crc32

The other three on-disk formats defined in terms of this checksum: the cache image self-checksum, the bisect session trailer, the undo journal trailer. Same swap, same values. check-in: 55d6e7c7bb user: dmestas tags: logstore-index

13:18
logstore: checksum through the vendored crc32

The four frame checksums plus the file header. The open-time scan CRCs every payload byte, so this is ~93% of a cold open: 416 MB goes 1.22s to 292ms. No format change — the values are identical by construction. check-in: dd71dd3006 user: dmestas tags: logstore-index

13:16
crc32: vendored slicing-by-8 CRC-32/ISO-HDLC

Bit-identical to std.hash.Crc32, which every on-disk checksum in the store is already defined in terms of. std's is byte-at-a-time over one table (~350 MB/s here); this folds eight bytes per step. Nothing uses it yet. check-in: aef8f539b5 user: dmestas tags: logstore-index

2026-07-28
22:58
rfcs: port applied errata; pin the §6.3 lowest-k tie-break and the one-name no-candidate refusal check-in: 307d7287fa user: dmestas tags: sub-manifest
18:29
submanifests: opt-in auto-retune dispatch and scalar /config propagation check-in: 4b882d36b8 user: dmestas tags: sub-manifest
16:43
submanifests: advise of drift at commit time, with an interactive retune offer check-in: 7b4e6df064 user: dmestas tags: sub-manifest
15:56
submanifests: add auto-retune setting and the §6.4 drift-check core check-in: 9c20a3746d user: dmestas tags: sub-manifest
11:49
submanifests: enable selects k by analysis instead of demanding one check-in: 081c3a0b90 user: dmestas tags: sub-manifest
More ↓