fix(tcfeed): surface requests that were asked and never sent a diff - #159
Merged
Conversation
The issue promises "a pull request is open alongside this with the diff". When that second half does not happen, nothing notices: `check` discovers its work by searching for open pull requests, so a repository holding a standing issue and no request is invisible to it permanently. james-6-23/codex2api sat in that state until the maintainer asked where the diff was, which is the wrong person to be running the check. `check` now also lists the repositories that were asked and never shown a diff, before the "no open requests" early return, since that is exactly the case where there is no pull request to report. Named, not repaired. `tcfeed pr <repo>` already finishes the pair on purpose, carrying the standing issue through so the request links back to it, and opening a request against somebody's repository should stay something typed rather than a side effect of asking for status. Also: the issue said "Two files under `.github/`" while openPr had been cut back to the workflow alone, so codex2api was asked on that sentence and then shown a one-file diff. Both now read the count off one INSTALLS list, so the question and the answer cannot disagree about the size of the change. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
ThreatCrush Security Scan63 finding(s) HIGH/CRITICAL: 4 | MEDIUM: 52 | LOW: 7
…and 13 more. Full results in the Security tab. Snippets are redacted; ThreatCrush never prints matched credential material. |
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Two defects, both found by a maintainer rather than by us.
james-6-23/codex2apiwas asked in issue #525 and never sent the diff. The issue text promises "a pull request is open alongside this with the diff", so that state is a promise made and then dropped. Nothing here noticed, and nothing would have:checkdiscovers its work by searching for open pull requests, so a repository holding a standing issue and no request is invisible to it permanently, not just for a while. The maintainer had to ask where the diff had got to.checknow also lists those repositories. It prints them before theno open requestsearly return, because a repository that was asked and never sent a diff is precisely the case where there is no pull request to report — the old order would have said "no open requests" and nothing about the promise outstanding.Named, not repaired.
tcfeed pr <repo>already finishes the pair deliberately, andprTargetcarries the standing issue through so the new request links back to it. Opening a request against somebody's repository should stay something a person types, not a side effect of asking for status.The second one is smaller and worse. The issue said:
openPrwas cut back to the workflow alone in #155 when the converter went away, butissueBodywas not, so every repository asked after that point was told "two files" and then shown a one-file diff. codex2api was asked on that sentence. Both now read the count off a singleINSTALLSlist, so the question and the answer cannot disagree about how big the change is.Verification
mastercopy of the same file: identical error set (nine classes, all missing@types/nodein a fresh worktree), so nothing new.head=ralyodio:threatcrush-scanit reports 0 unpaired (correct — codex2api was repaired by hand first); with a branch name that exists nowhere it reports 40. So it distinguishes present from absent rather than always returning empty.tcfeed checkruns end to end and prints no orphan block, which is the correct output today.tcfeed pr <repo> --dry-runrenders the issue asOne file under .github/against a one-file diff.Prettier is deliberately not run: it is not in
scriptsand not in CI, and running it reformatted 977 unrelated lines.