You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
fix(prototype): notes panel loses a row click when the list re-renders mid-press, flaking the browser suites #18090
The explore-directions browser suites flake on CI with a Playwright click timing out at 30 s, which turns main red and holds unattended merges until a newer main commit passes:
notes-edit-browser.test.sh, main run 44955798864 attempt 1, Test Bash (10): locator.click: Timeout 30000ms exceeded.
notes-region-browser.test.sh, main run 44968944394, Test Bash (0), commit 3d9d16a, which did not touch explore-directions: page.click: Timeout 30000ms exceeded on the reshape- button, then 6 dependent failures.
notes-copyout-browser.test.sh, PR run 44953095719: failed on a different symptom, the copy is said in the status line [""] (a 2 s status poll), not a click timeout.
Cause
render() in lib/ui-feedback/host.mjs (generated copy plugins/prototype/notes-bridge/host.mjs) rebuilds every notes row with replaceChildren on each server-sent state push and each named/view reply from the frame. When one lands between a row's mousedown and mouseup, the browser fires no click on the row, choose() never runs, and the row's actions (.acts, display: none until the row is current) stay hidden. The next click on Edit, Withdraw or Move or resize then waits 30 s for a button that never shows. The state stream polls every 0.3 s, and a contended CI leg widens the window. A forced render between mousedown and mouseup reproduces it every time.
Done when
A press on a notes row survives a re-render that lands mid-press: the row is selected and its actions show.
A regression assertion fails on the unfixed host and passes on the fixed one.
Failure counts before and after the fix are recorded for the forced-render probe and for a parallel run of the three suites under CPU contention.
Size: Small
Complexity: Medium
Uncertainty: Known
Flags: none
Blocked by: none
Blast radius: the notes host render/row-press handling in lib/ui-feedback/host.mjs, its generated copy under plugins/prototype/notes-bridge, and one new browser test; read by CI and anyone using the prototype notes panel.
Reversible: yes; revert the host change and the added test.
Open questions: none
Understood: a re-render landing between a notes row's mousedown and mouseup drops the click, so the row is never selected and its actions stay hidden. The next click then times out and turns main red intermittently. The refinement comment already holds the acceptance criteria (fix, regression assertion, regenerated copy, before/after failure counts). The report names a different symptom for the copyout suite, which the fix may not cover. No duplicate or merged fix found among the listed open issues and PRs. Related: #18042, #18016, #16723.
Not admitted to the issue lane: no trusted-human acceptance criteria: a listed human must put the confirmed - [ ] criteria in the body or post them as their own comment, or, on an issue a listed human filed, intake's newest refinement comment must hold them unedited. Admission takes the issue once each of those is met, when intake triages it again (.github/actions/resolve-config/admission-rules.json).
Refinement. A read-only run explored this repository, searched its issues and pull requests, and read the allowed docs to draft what follows. On an issue a listed human filed, it counts as the issue's spec once posted, while unedited; it is still model text, so check it.
Intended outcome: A press on a notes row in the notes host survives a re-render that lands between mousedown and mouseup, so the row is selected and its actions show, with a regression assertion that fails before the fix and passes after; the CI contention failure counts are a separate recorded measurement.
Acceptance criteria
lib/ui-feedback/host.mjs no longer loses a row press when render() runs between mousedown and mouseup: choose(entry) runs for that row (the row click listener near line 644 or render() at line 603 is changed so the press is not dropped).
After a render forced between a row's mousedown and mouseup, the row has aria-current=true and its .acts buttons (Edit, Withdraw, Move or resize where applicable) are visible.
plugins/prototype/notes-bridge/host.mjs is regenerated from lib/ui-feedback/host.mjs and matches the repo's generated-copy check.
A browser test under plugins/prototype/skills/explore-directions/scripts forces a render between mousedown and mouseup on a row and asserts the row is selected and its actions show.
That regression assertion fails when run against the unfixed host.mjs and passes against the fixed one, with both outcomes shown in the pull request description.
The forced-render probe failure count before and after the fix is recorded in the pull request description.
The failure counts before and after the fix for a parallel run of notes-edit-browser.test.sh, notes-region-browser.test.sh and notes-copyout-browser.test.sh under CPU contention are recorded in the pull request description.
The existing notes-*-browser.test.sh suites in plugins/prototype/skills/explore-directions/scripts still pass.
This records a re-fire of an earlier stop, not a new one. The start gate stopped run 44972603786, the work-resolve-issue run that agent-ready started, before any model turn, because the GITHUB_TOKEN budget was under its start-gate share, or the window was projected to have under its start-project share of the header's limit left by the run's expected end (github-token start-gate, start-project and start-run-minutes in rate-limits.json; the run's start-gate notice names which); that run ended at 2026-11-22T01:43:25Z. work-unblock has now removed and added agent-ready back to start a fresh run. A budget stop is the lanes' load, not a fault in this issue: it never counts toward limits.limit-stops-per-issue or the kill-switch wait.
Activity
github-actions commentedon Nov 22, 2026
Size: Small
Complexity: Medium
Uncertainty: Known
Flags: none
Blocked by: none
Blast radius: the notes host render/row-press handling in lib/ui-feedback/host.mjs, its generated copy under plugins/prototype/notes-bridge, and one new browser test; read by CI and anyone using the prototype notes panel.
Reversible: yes; revert the host change and the added test.
Open questions: none
Understood: a re-render landing between a notes row's mousedown and mouseup drops the click, so the row is never selected and its actions stay hidden. The next click then times out and turns main red intermittently. The refinement comment already holds the acceptance criteria (fix, regression assertion, regenerated copy, before/after failure counts). The report names a different symptom for the copyout suite, which the fix may not cover. No duplicate or merged fix found among the listed open issues and PRs. Related: #18042, #18016, #16723.
Automated triage. A maintainer reviews it.
melodic-automation-lanes commentedon Nov 22, 2026
Not admitted to the issue lane: no trusted-human acceptance criteria: a listed human must put the confirmed
- [ ]criteria in the body or post them as their own comment, or, on an issue a listed human filed, intake's newest refinement comment must hold them unedited. Admission takes the issue once each of those is met, when intake triages it again (.github/actions/resolve-config/admission-rules.json).melodic-automation-lanes commentedon Nov 22, 2026
Refinement. A read-only run explored this repository, searched its issues and pull requests, and read the allowed docs to draft what follows. On an issue a listed human filed, it counts as the issue's spec once posted, while unedited; it is still model text, so check it.
Intended outcome: A press on a notes row in the notes host survives a re-render that lands between mousedown and mouseup, so the row is selected and its actions show, with a regression assertion that fails before the fix and passes after; the CI contention failure counts are a separate recorded measurement.
Acceptance criteria
External state: none
Related work
Sources read
Automated refinement by the lanes App. Intake triages the issue again with it.
melodic-automation-lanes commentedon Nov 22, 2026
{"v":1,"actor_id":337595936,"title_sha256":"09a9fcecf3c84d2496b12dfad0be6efb75cea0a20898f3ce195aaf1b929b605f","body_sha256":"b758cdbddde1947f2f914da5c17d94317bbde0b39eefcde17d8e5f70a7662272"}
melodic-automation-lanes commentedon Nov 22, 2026
This records a re-fire of an earlier stop, not a new one. The start gate stopped run 44972603786, the work-resolve-issue run that
agent-readystarted, before any model turn, because the GITHUB_TOKEN budget was under its start-gate share, or the window was projected to have under its start-project share of the header's limit left by the run's expected end (github-token start-gate, start-project and start-run-minutes in rate-limits.json; the run's start-gate notice names which); that run ended at 2026-11-22T01:43:25Z. work-unblock has now removed and addedagent-readyback to start a fresh run. A budget stop is the lanes' load, not a fault in this issue: it never counts toward limits.limit-stops-per-issue or thekill-switchwait.github-actions commentedon Nov 22, 2026