10 most recent check-ins occurring on or after 2026-09-27 20:37:13.
More ↑
|
2026-09-28
| ||
| 20:37 | commit holds the store lock only to snapshot and to publish, not through the tree scan, editor or signing check-in: 6145d7a7b8 user: dmestas tags: trunk | |
| 20:22 | A writer turned away by a busy lock says so once; serve answers it 503 check-in: 5a0b10fa8d user: dmestas tags: trunk | |
| 19:37 | Lock notices append to a redirected stderr instead of overwriting it check-in: d59b4a1827 user: dmestas tags: trunk | |
| 16:44 |
zit git export/status: a remembered Git mirror with autopush
git export exports into MIRROR and remembers it, with an autopush URL, push branch and email domain, in repository config; a bare re-run does the round trip. Authors' addresses come from their user-table contact, set with the new users contact. Autopush URLs carrying a password are refused. The config payload bound is now sized from its key count. Release 0.3.34. check-in: 5fce7e867d user: dmestas tags: trunk, version-0.3.34 | |
| 12:53 |
pack, shun --force: fsync the directory after the rename; model it in DST
The rename that installs a new generation was never made durable, so a power loss after a reported success could bring the old store back, and with it content a shun had removed. The fault layer now reverts creates, renames and deletes no directory fsync made durable, and the shun sweep requires a rewrite that succeeded to survive a power cut. check-in: 3873759cfa user: dmestas tags: trunk | |
| 12:35 |
landing accept: read the target once; --expect-target refuses a moved one
accept merged onto the target resolved by name at merge time, so a landing accepted meanwhile put an untested target under the merge. It now reads the target once and merges onto that check-in by hash. --expect-target HASH refuses with exit 7 and nothing written when the target is no longer at HASH. check-in: 676068f793 user: dmestas tags: trunk | |
| 11:59 |
Lock the store with exclusive-create lock files instead of flock
Writers create .zit.lock with O_CREAT|O_EXCL before opening the store. A waiter names the holder and gives up after a minute; a dead holder's lock is broken; pack and shun --force hold the writer lock across their final check and rename. Release 0.3.33. check-in: d6cf7bf59c user: dmestas tags: trunk, version-0.3.33 | |
| 02:59 |
import-git: record a project code; verify notes a store without one
import-git wrote no project code, so every sync peer and forge refused the store it made. It now derives one from the destination's absolute path, as init does, and the import request requires it. verify notes a repository with no project code without counting it as damage. check-in: add51a12d2 user: dmestas tags: trunk | |
| 02:39 |
Take the writer lock inside LogStore.open and test the lock class
LogStore.open now takes the exclusive lock before it reads the file's length and scans for the end, and a read-only handle refuses to append. purge undo opens the store through the shared writer path. A new DST sweep runs two to four writers interleaved against one log through a file that models the lock. It flags unlocked or misplaced appends as they happen, checks each writer sees everything committed before it locked and reads back its own writes, including deltas. A source check fails on any write-mode open of the store outside openStoreAppend. An e2e test races commits, tags and user-table writes on one store and counts what landed. check-in: 5c445bb3f3 user: dmestas tags: trunk, version-0.3.32 | |
|
2026-09-27
| ||
| 20:37 |
purge undo: take the store's append lock before restoring
It appended without the lock every other writer takes, so a concurrent commit, pull or push could overwrite its frames or have its own overwritten with valid frames nothing could detect. It now opens through openStoreAppend, which also resolves a detached worktree's store. Release 0.3.32. check-in: 2249f19056 user: dmestas tags: trunk | |