Repository navigation
fix(issue-radar): keep focus on the Refresh issues buttons while they refresh - #31481
Conversation
UX Review (Opus 5.5) — ✅ PASSUX-level review of UX-Verdict: PASS Nothing blocks: the buttons look the same as before, and the only change is that keyboard focus stays on them during a refresh. With no visual change, the missing blind read leaves no control or state unreviewed. Notes
[UX-REVIEWED] 49dbaed Description read[DESCRIPTION-READ] 0e93a32e750ce5eff8c4cb06e6e4f631402604bd582e25c849ce8ff6763314b2 That sha256 names everything this verdict read: the PR description, plus the 3 evidence file(s) the workflow collected from it and handed this review. It covers that evidence because the media strip replaces every attachment URL with the same placeholder, so an edit swapping one attachment for another would leave the description's captured text identical while what this review judged changed. Recompute it with |
Design Review (Opus 5.5) — ✅ PASSDesign-level review of Design-Verdict: PASS Nothing blocks: the fix uses the same aria-disabled pattern as #31360 and #31184, Issue Radar has no adoption pin, and the change takes nothing away. Notes
[DESIGN-REVIEWED] 49dbaed Description read[DESCRIPTION-READ] f8a881376513088400dd9b813dcb278afaf844bcb6dd1e1488524853ed3e2a33 That sha256 names everything this verdict read: the PR description, plus the 1 evidence file(s) the workflow collected from it and handed this review. It covers that evidence because the media strip replaces every attachment URL with the same placeholder, so an edit swapping one attachment for another would leave the description's captured text identical while what this review judged changed. Recompute it with |
First Principles Review (Opus 5.5) — ✅ PASSPremise-level review of First-Principles-Verdict: PASS Nothing blocks: two buttons with the same reported defect (#31479) get the inert-but-focusable shape the repo already uses. The siblings are counted and deferred to #31618. What this change shipsInventory (4 items) — 4 justifiedIntent: A keyboard or screen-reader user keeps focus on Issue Radar's Refresh issues button while a refresh runs. This is a FIX.
[FIRST-PRINCIPLES-REVIEWED] 49dbaed Description read[DESCRIPTION-READ] d6236652384aae1a2c4cf27aaff3c42437bbc06bacfd255e0167f1c59bfbbdd7 That sha256 names the PR description this verdict read. Recompute it with |
Opus 5.5 Review — ✅ no blocking findingsReviewed Verdict parsed from the review's SHA-scoped output markers for commit False positive or not applicable? A repository writer can comment: |
GPT 6.1 Review — ✅ no blocking findingsGPT 6.1 completed its review of This comment is updated in place on each push. Review detailsNo findings. False positive or not applicable? A repository writer can comment: |
6dfc1ae to
cb0264e
Compare
iamwhatever
left a comment
There was a problem hiding this comment.
Requesting changes for one missing follow-up issue. The code is sound.
Reviewed head: cb0264e
The fix is right. aria-disabled plus the if (!refreshing) click guard keeps focus on the button and still ignores a press while busy (IssueList.tsx#L233, OverviewView.tsx#L147). The new tests would fail on main. All checks on this head are green.
Blocker: deferred siblings have no linked issue. Not a goal says the other Issue Radar refresh controls are "separate items", and First Principles found one more the description does not mention. The same disabled={...refreshing} focus drop remains at:
- PrList.tsx#L378 (
disabled={pullsRefreshing}, not mentioned) - GraphView.tsx#L882 and #L964
- PipelineView.tsx#L232
#31479 covers only the two buttons fixed here. Please either fix these in this PR or open an issue covering the remaining Issue Radar refresh buttons (the list above plus the CI runs, group reviews and detail-menu controls you named) and link it in the description.
AI findings judged non-blocking: Design note on the default cursor while busy. It is an intentional aria-disabled:cursor-default and the only visible change.
— crew-pr-reviewer (r148)
|
@iamwhatever Thanks. I filed the remaining Issue Radar sites as #31618: PrList.tsx:378, GraphView.tsx:882 and :964, PipelineView.tsx:232, plus the other fetch-disabled controls in that app. Not a goal now says "Deferred: #31618", and the sibling sweep lists each one. No code change. |
cb0264e to
49dbaed
Compare
|
Since the review on cb0264e:
|
bolichen97
left a comment
There was a problem hiding this comment.
Approving head 49dbaed. The gate shows 97 checks passed, none red or pending.
- Both Refresh issues buttons (
IssueList,OverviewView) swap the nativedisabledforaria-disabledplus anif (!refreshing)guard inonClick. The button keeps keyboard focus during a refresh, a press while busy does nothing, and the busy name and title still say why. The new tests focus the button, fliprefreshing, and assert focus stays and a busy click does not callrefresh. - The styling moves to
aria-disabled:variants, so the busy look is unchanged.
@iamwhatever could you take another look? The one blocker in your change request on cb0264e, a linked issue for the deferred sibling controls, is addressed: #31618 lists PrList.tsx:378, GraphView.tsx:882 and :964, and PipelineView.tsx:232, and Not a goal now reads "Deferred: #31618". The code is unchanged since your review apart from a rebase.
iamwhatever
left a comment
There was a problem hiding this comment.
Approve at 49dbaed.
Both Refresh issues buttons now use aria-disabled plus an if (!refreshing) click guard instead of native disabled. The button keeps keyboard focus during a refresh, and a press while busy still does nothing (IssueList.tsx#L233-L241, OverviewView.tsx#L147-L155). The new tests focus the button, flip refreshing, and check focus is kept and a busy click is ignored; they would fail on main.
The blocker in our r148 change request is cleared: #31618 lists every remaining Issue Radar control with the same focus drop, and Not a goal links it.
AI findings judged non-blocking:
- UX note: the busy opacity also fades the focus ring. Before this PR focus left the button entirely, and main already uses the same
aria-disabled:opacity-*shape on the siblings fixed by #31687 (IssueList.tsx#L241). - Design note: a shared busy-button helper would keep the attribute and the guard together. This is a refactor idea; both sites set both parts (OverviewView.tsx#L147-L155).
CI: the only red check is Sensitive Change Review attempt 3, which hit a GitHub API rate limit after the approval re-run; attempt 2 on this head passed and the required PR Readiness is green. Re-running that job would clear the row. I read the source and CI; I did not run the tests.
— crew-pr-reviewer (r149)
Red check Sensitive Change Review (run 42380271030) does not count: its log ends with "Could not read the PR from the API (CalledProcessError: gh: API rate limit exceeded for installation ...)", an API rate-limit error, not a finding.
…or notice Issue Radar's Refresh ran through a mutation whose error nothing read, so a failed refresh just stopped the spinner and left stale rows as if it had worked. Its failure now reaches issuesError, which IssueList already shows in an ErrorNotice, the way pullsError reports a failed PR refresh. BuiltinAppRoute's release lookup kept only the query's data, so a lookup that kept failing showed nothing while the compiled page rendered against a backend that may not be there. Once its retries are spent the error shows in an ErrorNotice above the page; the page itself still renders. Refs #30664, #31481
…or notice (#33029) * fix(ui): show a failed issue refresh and app release lookup in an error notice Issue Radar's Refresh ran through a mutation whose error nothing read, so a failed refresh just stopped the spinner and left stale rows as if it had worked. Its failure now reaches issuesError, which IssueList already shows in an ErrorNotice, the way pullsError reports a failed PR refresh. BuiltinAppRoute's release lookup kept only the query's data, so a lookup that kept failing showed nothing while the compiled page rendered against a backend that may not be there. Once its retries are spent the error shows in an ErrorNotice above the page; the page itself still renders. Refs #30664, #31481 * test(ui): expect the release lookup notice after a failed lookup The legacy-route test pinned no alert after a failed release lookup, which was the silent behaviour this PR replaces with an ErrorNotice.
…or notice (#33029) * fix(ui): show a failed issue refresh and app release lookup in an error notice Issue Radar's Refresh ran through a mutation whose error nothing read, so a failed refresh just stopped the spinner and left stale rows as if it had worked. Its failure now reaches issuesError, which IssueList already shows in an ErrorNotice, the way pullsError reports a failed PR refresh. BuiltinAppRoute's release lookup kept only the query's data, so a lookup that kept failing showed nothing while the compiled page rendered against a backend that may not be there. Once its retries are spent the error shows in an ErrorNotice above the page; the page itself still renders. Refs #30664, #31481 * test(ui): expect the release lookup notice after a failed lookup The legacy-route test pinned no alert after a failed release lookup, which was the silent behaviour this PR replaces with an ErrorNotice. Release resolution: docs/system-specs/modules/app-kit-platform.md keeps release's surrounding text; only the failed-lookup sentence changes. Release resolution: BuiltinAppRoute.tsx adds the ErrorNotice import, which main got from #30667 (not on release; too wide to pick for one import). (cherry picked from commit e43f72c)
…or notice (#33029) * fix(ui): show a failed issue refresh and app release lookup in an error notice Issue Radar's Refresh ran through a mutation whose error nothing read, so a failed refresh just stopped the spinner and left stale rows as if it had worked. Its failure now reaches issuesError, which IssueList already shows in an ErrorNotice, the way pullsError reports a failed PR refresh. BuiltinAppRoute's release lookup kept only the query's data, so a lookup that kept failing showed nothing while the compiled page rendered against a backend that may not be there. Once its retries are spent the error shows in an ErrorNotice above the page; the page itself still renders. Refs #30664, #31481 * test(ui): expect the release lookup notice after a failed lookup The legacy-route test pinned no alert after a failed release lookup, which was the silent behaviour this PR replaces with an ErrorNotice. Release resolution: docs/system-specs/modules/app-kit-platform.md keeps release's surrounding text; only the failed-lookup sentence changes. Release resolution: BuiltinAppRoute.tsx adds the ErrorNotice import, which main got from #30667 (not on release; too wide to pick for one import). (cherry picked from commit e43f72c)
Problem / Motivation
Goal: Keyboard focus stays on Issue Radar's Refresh issues buttons while a refresh runs.
In
website/src/apps/issue-radar/views/OverviewView.tsxandwebsite/src/apps/issue-radar/components/IssueList.tsx, the Refresh issues button sets the nativedisabledattribute whilerefreshing. A disabled button leaves the tab order, so a keyboard press drops focus to the page for the whole refresh, and it never comes back.Why it matters
A keyboard or screen-reader user loses their place in Issue Radar each time they refresh the issue list.
Not a goal
What changed (motivation → approach → change)
Symptom: focus falls to the page on a keyboard refresh. Cause:
disabled={refreshing}takes the focused button out of the tab order. Change: the same inert-but-focusable shape as the chat PR and issue panels (#31360) and the Discover store Refresh (#31184):aria-disabled={refreshing || undefined}witharia-disabled:classes in place of thedisabled:ones, andonClick={() => { if (!refreshing) refresh() }}so a press while busy still does nothing, as the nativedisableddid.Backwards compatibility
Compatible: the buttons, their names and the refresh they run are unchanged; a press while busy is still ignored, now by the click guard.
Removes nothing: no route, key or behaviour is taken away.
Tests
website/src/apps/issue-radar/views/OverviewView.coverage.test.tsxandwebsite/src/apps/issue-radar/components/IssueList.refreshButton.test.tsx: each gains a case where focus stays on Refresh while busy, a press while busy does not callrefresh, and focus is still there after it ends. Each busy-name case now assertsaria-disabledinstead ofdisabled.All four fail on main and pass here.
Proof: not provable -- frontend test (prove.py runs Python only)
vitest run src/apps/issue-radar: 1643 passed.tsc -p tsconfig.app.jsonand eslint: clean. Local GPT 5.6 Sol and Opus 5.5 reviews: no findings.Manual verification
N/A — unit coverage sufficient: the tests drive the mounted views.
Screenshots / video
Why no screenshot: nothing on screen changes; only whether keyboard focus stays on the button.
Related Issues
Fixes #31479
Sibling sweep
Swept: git grep -niE "disabled={[^}]*(refreshing|isfetching)" -- website/src/apps/issue-radar (run on origin/main)
Sibling: website/src/apps/issue-radar/views/OverviewView.tsx -- changed here
Sibling: website/src/apps/issue-radar/components/IssueList.tsx -- changed here
Sibling: website/src/apps/issue-radar/components/PrList.tsx -- tracked in #31618
Sibling: website/src/apps/issue-radar/views/GraphView.tsx -- tracked in #31618
Sibling: website/src/apps/issue-radar/pipeline/views/PipelineView.tsx -- tracked in #31618
Sibling: website/src/apps/issue-radar/views/CiRunsView.tsx -- tracked in #31618
Sibling: website/src/apps/issue-radar/views/GroupReviewsView.tsx -- tracked in #31618
Sibling: website/src/apps/issue-radar/components/IssueDetail.tsx -- tracked in #31618
Sibling: website/src/apps/issue-radar/components/PrDetail.tsx -- tracked in #31618
Pattern harvest
Rule candidate: none
Pattern: a busy button that sets native
disabledon itself while focused.Checklist
feat|fix|docs|style|refactor|perf|test|chore|ci|build|revert: ...)