Repository navigation
Fix three intermittent CI failures, one a sync race at quit - #38
Merged
MHohlios merged 2 commits intoNov 4, 2026
Merged
Conversation
added 2 commits
November 3, 2026 16:07
A network step given up on at its limit leaves its git thread running, still holding git's locks. A sync quit during a fetch or push started the quit's own fetch or push beside it, which failed on the tracking branch's lock or the remote's ref, so the last sync at quit was lost. Each network step on a repo now waits for the last one's thread to end, within its own limit. Two tests raced themselves: - The restore test saved a map version through a second writer before the engine's writer had put room 1 on the disk, so the version could be empty. It now flushes the engine first. - The hanging helper test started its 200 ms limit before the shell helper ran, so a slow start was killed before it wrote its pid. The limit now ends once the helper is running.
MHohlios
deleted the
wiffletree/fix-three-intermittent-test-failures-in-ci
branch
November 4, 2026 00:01
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Three tests failed intermittently on GitHub's Ubuntu runner, which would also stop a release. Two were races in the tests. One was a real race in sync: quitting mid-sync could lose the final commit and push. Linear: VYT-263.
Causes and fixes
sync::tests::a_folder_that_was_the_player_s_own_and_changed_by_sync_waits_again). A quit gives up on the running pass's fetch or push, but its git thread runs on, holding git's locks. The quit's own final pass then fetched beside it and failed with a lock error, shown as "The repo is busy".Repo::off_threadnow waits, within its own limit, for a given-up network step's thread to end before starting the next one. Quit is still bounded byquit_wait. The new testgit::tests::a_network_step_never_runs_beside_one_given_up_onfails every time without the fix and passes with it.session::tests::a_restore_waits_out_a_merge_landing). The test saved the map version through a second writer before the engine's writer had written room 1, so it could save an empty map. It now callsengine.flush()first. In the app, versions go through the engine's own writer, in order.git::credentials::tests::a_hanging_helper_is_killed_when_the_limit_is_up). The 200 ms limit could end before the shell ran. The limit is now raised by a watcher once the helper has written its pid. The test still checks that the helper is killed and gone.Checks
cargo fmt --check, clippy (--release --locked -D warnings) andcargo test --workspace --no-fail-fast --lockedpass.Follow-ups
crates/mudling-core/src/git/mod.rs:59).crates/mudling-core/src/git/mod.rs:324).Reviewed with Wiffletree: 1 full review and 1 re-check, 2 rounds (Tests, Claude, Style)