Repository navigation
Conversation
…ntu 26.04 fix Every review round since v0.45.0 found a new two-simultaneous-holders interleaving in the lock takeover/recovery features added in #1106, #1120 and #1123 (see #1119, #1122, #1124/#1125, PRs #1126 and #1129). Rather than chase another interleaving, revert e704ea2 (#1101/#1103), 4f42987 (#1119/#1120) and 2fd0547 (#1122/#1123) in full and keep only 6ca944a (#1108/#1112, the noclobber-guarded pid write on the primary mkdir path) -- the lock as shipped in v0.45.0, which users ran for weeks without a two-holder report, plus the Ubuntu 26.04 fix. scripts/lib-lock.sh and tests/test_lock_primitive.py now match v0.45.0 plus exactly the #1108/#1112 hunks (diffed and confirmed against both). tests/test_lock_fork_reduction_1056.py is restored to its pre-#1106 content (it tested fork-count regressions in the now-removed code). The comments in scripts/save-session.sh describing the removed takeover are reverted to their v0.45.0 wording. changelog.d/1119.fixed.md and changelog.d/1122.fixed.md are removed since they describe behavior that no longer ships; the dangling trap.d fragments about the removed code (1101-1103, 1122) go with them. Added changelog.d/1131.changed.md explaining the rollback in plain terms and noting that #1101 and #1103 are reopened. Redesigning the stuck-lock takeover/recovery feature is deferred to a separate pull request with more review time. Maintainer-approved 2026-11-14. Closes #1131 Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
…both dropped settings and the reopened #1103 hang Co-Authored-By: Max <noreply>
…ack actually ships Co-Authored-By: Max <noreply>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Closes #1131
Why
Since v0.45.0 the lock took on takeover and recovery features (#1106 for #1101/#1103, then #1120 and #1123). Every review round since found a new way for two processes to hold the same lock at once (#1119, #1122, #1124/#1125, and the reviews on PRs #1126 and #1129, neither merged). v0.45.1 ships one of those bugs. The maintainer decided (2026-11-14) to put the lock back exactly as v0.45.0 had it, which users ran for weeks with no two-holder report, keep only the #1108 Ubuntu 26.04 fix, and redesign recovery separately.
What changed
scripts/lib-lock.sh= v0.45.0 + Fix #1108: lock_acquire no longer trusts mkdir's exit code alone on Ubuntu 26.04 #1112's hunks, nothing else. Verified: the diff from v0.45.0 matches Fix #1108: lock_acquire no longer trusts mkdir's exit code alone on Ubuntu 26.04 #1112's own lib-lock.sh diff line for line.tests/test_lock_primitive.py= v0.45.0 + Fix #1108: lock_acquire no longer trusts mkdir's exit code alone on Ubuntu 26.04 #1112's tests (+229 lines vs v0.45.0).tests/test_lock_fork_reduction_1056.pyandscripts/save-session.share identical to v0.45.0.changelog.d/1131.changed.mdexplains the rollback in plain words. Two trap.d notes about the removed code are removed.claude -pinherits CLAUDE_CODE_EFFORT_LEVEL: at max, NDC and consolidation never finish inside their timeouts #1128 (already on main).#1101 and #1103 will be reopened after merge, with an explanation to their reporters.
[AI-generated]