Repository navigation
π fix(park-recheck): a park with options waits on what the options name - #16840
Conversation
projectbluefin/server#614 parked on hivecommons#583 (both options merge it) and listed hivecommons#584 and hivecommons#588 among the PRs its red prow check touched. The re-check reported those two merging as "the park reason names it" and called the park out of date while hivecommons#583 was still open. When a park's lettered options (each option line and the lines under it) name references, only those count as named refs; the rest of the text is evidence. Parks without options, or whose options name nothing, keep every named reference. PRs that refer to the issue still count as before. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BQYnPGFJx81c8JXR3o1sg1 Signed-off-by: Danathar <Danathar@users.noreply.github.com>
|
[APPROVALNOTIFIER] This PR is NOT APPROVED This pull-request has been approved by: The full list of commands accepted by this bot can be found here. DetailsNeeds approval from an approver in each of these files:Approvers can indicate their approval by writing |
Open PR overlap checkBase branch: Conflicts are checked with every PR merged onto the current tip of No conflicting open PRs or same-file overlaps were found. |
There was a problem hiding this comment.
Read against v6 with the callers. Looks correct to me.
parkOptionRefsreusesunparkOptionLetter(src/pkg/github/issue_unpark_command.go:1580), which onv6already accepts theOption A:,**A.**bold and plainA)forms, so the option-block boundaries match what the "What to reply" menu sees.- The substitution at
src/pkg/github/issue_park_recheck.go:181-183only narrowsnamed; the "PR refers to the issue" branch inparkChanges(issue_park_recheck.go:493) is untouched, so a mergedRefs #614PR still produces the note. The new test's collateral case passes because #584/#588 are cross-refs whose bodies don't refer to #614, not because that path was disabled. - Falling back to every named ref when options name nothing (
len(acted) > 0guard) keeps the pre-existing behaviour for parks without options.
The trade-off the body names (prose waits on #X, options name #Y β only #Y watched) is real but documented in maintainer-commands.md; fine by me as an engineering call.
β hive: agent=reviewer-2 backend=copilot model=claude-fable-5.1 copilot=1.0.88
Executive summary
On 11-03 the park re-check told maintainers that projectbluefin/server#614's park "may be out of date". It listed #588 and #584 as merged "(the park reason names it)". Both were collateral: the park was about #583, which both of its options merge and which is still open. #584 and #588 appear only in the evidence line "red on
main(cron) and on PRs #583, #584, #588, #594, #613".With this PR, when a park's lettered options name references, only those count as what the park waits on. The rest of the park text is evidence.
What
#Nin the park text#583)#N(unchanged)PRs that refer to the issue ("Refs #614", closing keywords, β¦) and merge still produce the note as before.
How
parkOptionRefs(src/pkg/github/issue_park_recheck.go) collects each option line (unparkOptionLetter, the same parser the "What to reply" menu uses) and the lines under it, up to a blank line or the next option. It then runsparkNamedRefsover those blocks.recheckParkedIssueuses that list fornamedwhen it is non-empty. That affects the lookups,parkChanges, the adopted-copy follow-up andmarkParkLandings.Tests
TestParkRecheckOptionsNameWhatTheParkWaitsOnTestParkRecheckOptionsWithoutRefsKeepNamedRefsTestParkOptionRefs#583; continuation lines count; a blank line ends a block; no options β nilTestParkRecheck*go test ./pkg/github/ ./pkg/writeback/check-docs-citations.py src/docsNotes
π€ Generated with Claude Code
https://claude.ai/code/session_01BQYnPGFJx81c8JXR3o1sg1