Skip to content

Fork: bring in review.prComment (no PR review comments) - #13

Merged
jbutton merged 4 commits into
mainfrom
work/24-fork-bring-in-review-prcomment
Nov 12, 2026
Merged

jbutton merged 4 commits into
mainfrom
work/24-fork-bring-in-review-prcomment

Conversation

@jbutton

@jbutton jbutton commented Nov 12, 2026

Copy link
Copy Markdown
Owner

The review comment on pull requests can now be turned off: review.prComment is on this branch (commit ce91f714), ported from 809e63a4. The full npm run check -- --all passes (1330 tests).

  • Default: still on, as upstream has it. With it off, the board posts no review comment on the PR. The review still runs, is judged at the gate, and shows on the review page and in the task panel.

  • Conflicts: main had dropped pr_number(...) from the one gh pr comment call, so I kept main's version and added the switch. Two settings-page tests needed their screen height raised by one row for the new setting (65→66 and 66→67).

  • One change from the old commit: the setting is now read when the review ends, not when it starts. A review can run 20 minutes, so turning it off during one already stops that comment.

  • Every path is covered: all reviews that post a comment today (first and later rounds, failed and passed, and the chosen attempt's review) go through that one call in crates/flow/src/finish/pipeline_end.rs.

  • Tests (tests/gate_more.rs): the old test is kept. Two new tests run the same task with the switch on and off and compare every gh call the board makes:

    • a review that fails, then a second round that passes;
    • attempts, where a reviewer picks one and it goes on to review.

    The only difference between on and off is the review comments, so nothing else the board writes to the PR changes. I also removed the switch from the code as a test, and all three tests failed.

  • Docs: docs/gate.md and docs/settings.md (the settings table was regenerated by the check).

No version bump, tags, push or anything sent to eliasstravik/*.

Also noticed (not touched):

  • tests/coordinator.rs, tests/coordinator_chat.rs, tests/coordinator_chat_more.rs: the compiler warns about an unused variable top.
  • tests/board_list/pinned.rs:230: the compiler warns about a .clone() that does nothing.

Task: PRJ-24 (herdr-projects/24)

jbutton and others added 4 commits November 12, 2026 23:44
Brought over from 809e63a4 (24 Oct, never merged) onto today's main. The
default stays on, as upstream has it; turned off, the board posts no
review comment on the task's pull request, and the review still runs, is
judged at the gate and shows on the review page and in the task panel.

The switch is read as each review ends, not when it starts, so turning it
off during a review already counts. Every review that comments today (the
first and later rounds, failed and passed, the chosen attempt's) goes
through the one call in pipeline_end.rs.

Tests run the same task with it on and off and check that the gh calls
differ only by the review comments: nothing else the board writes to the
pull request changes.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@jbutton
jbutton marked this pull request as ready for review November 12, 2026 23:05
@jbutton
jbutton merged commit c7eeab9 into main Nov 12, 2026
0 of 4 checks passed
@jbutton
jbutton deleted the work/24-fork-bring-in-review-prcomment branch November 12, 2026 23:06
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant