docs: correct a misattributed ledger citation on the refresh livelock, and record the fix - #16
Merged
Merged
Conversation
…, and record the fix (Subject deliberately avoids the literal "BACKLOG #N" token: the claim gate reads that in a code-touching commit's SUBJECT as "this commit implements item N" and demands the number be claimed first. This commit REMOVES a wrong citation rather than implementing anything, so claiming would be the wrong signal. The body below names the item freely.) BACKLOG #17 is the py3.11 pytest/aiosqlite cancellation deadlock, closed OBSOLETE at the 3.14-only migration. It is cited correctly in tee/relay.py, the webconsole conftest and test_smart_backend.py. It is NOT the harness message-list livelock. A pre-existing comment on the flaky monitor test blamed "#17" for that, and while fixing the livelock I copied the wrong citation into four more places -- two test docstrings, an assertion message and the push-guard header -- plus the commit messages and PR bodies that went with them. Anyone following it lands on a closed item about an unrelated bug, which is a slower and more confusing dead end than no citation at all. Removed from the livelock sites; each now says plainly what the defect was, and notes what the wrong citation used to claim so the next person does not re-derive it. The legitimate #17 references are untouched. Also records the livelock itself in CHANGELOG 0.3.1, where it shipped: PR #5 merged before the tag was cut, but no changelog line went with it, so the release notes are silent on a user-visible fix to the harness message list. Filed under Fixed with the measurement that identified it (391 reads served, 0 rendered) rather than a vague "fixed a refresh bug". And corrects the push guard's rationale for leaving enforce_admins off. It cited the "known flaky test" as the reason an admin override must stay available; that failure was a livelock, not a flake, and it is fixed -- so the argument is weaker now than when it was written. Said so rather than leaving a stale justification standing for a security-relevant setting.
wshallwshall
enabled auto-merge (squash)
July 27, 2026 21:32
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
1 participant
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.
Item 17 in the ledger is the py3.11 pytest/aiosqlite cancellation deadlock, closed OBSOLETE at the 3.14-only migration. It's cited correctly in
tee/relay.py, the webconsoleconftest.py, andtest_smart_backend.py— those are untouched.It is not the harness message-list livelock. A pre-existing comment on the flaky monitor test blamed item 17 for that, and while fixing the livelock I copied the wrong citation into four more places — two test docstrings, an assertion message, and the push-guard header — plus the commit messages and PR bodies that carried them.
Following that citation lands you on a closed item about a different bug, which is a slower dead end than no citation at all. It's the same doc-rot shape as the release guard and the
dependabot-lock-resynccomment: a statement that was true once, propagated by copying rather than checking. I propagated this one.Each site now says plainly what the defect was, and notes what the wrong citation used to claim so nobody re-derives it.
Also
The livelock is recorded in CHANGELOG 0.3.1, where it actually shipped. #5 merged before the tag was cut but carried no changelog line, so the release notes were silent on a user-visible fix to the harness message list. Filed under Fixed with the measurement that identified it — 391 reads served, 0 rendered — rather than a vague "fixed a refresh bug".
The push guard's rationale is corrected. It cited the "known flaky test" as the reason
enforce_adminsmust stay off so an admin override remains available. That failure was a livelock, not a flake, and it's fixed — so the argument is weaker now than when it was written. Better to say so than leave a stale justification standing behind a security-relevant setting.Note on the commit subject
It deliberately avoids the literal
BACKLOG #Ntoken. The claim gate reads that in a code-touching commit's subject as "this commit implements item N" and requires the number be claimed first — correct behaviour, wrong signal here, since this removes a citation rather than implementing anything.🤖 Generated with Claude Code