Repository navigation
chore: remove the feature-preview one-time carry-up - #32359
Conversation
UX Review (Opus 5.5) — ✅ PASSUX-level review of UX-Verdict: PASS Nothing to check: the diff only deletes two notices about saving a switch to the gateway and adds or changes no control, so no screenshot is needed. Notes
[UX-REVIEWED] 38f63cf Description read[DESCRIPTION-READ] 6db093c707bae556c4631ae2548c03ec4881a1e83a1e128346aeabda3d2510c2 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: it removes a migration whose stable release has shipped. The one break, a browser-only ON that was never synced now reading as off, is listed and accepted, and a user can undo it by turning the switch back on. Notes
[DESIGN-REVIEWED] 38f63cf Description read[DESCRIPTION-READ] 0b18f3b41974bf8b4e3a24920cf2ef230e86d8de0008947963acbf0e4025e0ca 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: the change deletes a one-time migration whose release has shipped (#24963). The one addition, clearing two dead localStorage keys on mount, protects against nothing. Notes
What this change shipsInventory (6 items) — 5 justifiedIntent: let each Feature previews switch take its value from the gateway alone, now that the migration has done its job. This is a removal (ADDITION-free chore), not a FIX.
[FIRST-PRINCIPLES-REVIEWED] 38f63cf Description read[DESCRIPTION-READ] 02d191b789f3f9c6a306b1afdf6c1b4eb849a378b21a6892f94ccf0a1fd772f1 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: |
73d1902 to
a2b453d
Compare
|
Rebased: main moved in docs/system-specs/modules/learn-cron-dashboard.md, website/src/pages/settings/FeaturePreviewsSection.tsx and the website/src/i18n/locales/*.json files since this head went green (f58c3cf -> e356cc6); CI re-running. The rebase was clean and the patch is unchanged. The 47 frontend test files that cover the preview flags and Feature Previews section pass on the new base (1079 tests), and i18n:check, lint:i18n, eslint and the app typecheck pass too. [operator: bolichen97] |
v0.12.0 shipped #22913, which keeps the Settings > Developer > Feature previews switches on the gateway. Its migration for browsers that set a switch before then (an if_unset carry-up of a browser-only ON, the kc-feature-previews-synced marker and the per-switch server-backed record) is no longer needed, so the gateway value is now the only source of truth and localStorage a pure cache: a switch the gateway holds no value for reads as off, and the two retired keys are cleared when the sync mounts. The carry-refused notices and their strings go with it in every catalog. Reconcile, in-flight protection and refused-toggle revert are unchanged.
a2b453d to
38f63cf
Compare
|
Rebased: main moved in docs/system-specs/modules/learn-cron-dashboard.md and the website/src/i18n/locales/*.json files since this head went green (e356cc6 -> 21597a3); CI re-running. The rebase was clean and the patch is unchanged. The previous head's red checks were all GitHub API rate-limit failures inside CI (no review finding, no test failure), so this run replaces them. The 48 frontend test files that cover the preview flags and the Feature Previews section pass on the new base (1102 tests), and i18n:check, lint:i18n, eslint and the app typecheck pass too. [operator: bolichen97] |
KiroCrewCaptain
left a comment
There was a problem hiding this comment.
Approved: PR Readiness passed on 38f63cf, and all five review lanes (Opus 5.5, GPT 6.1, Design, UX, First Principles) report no blocking finding on this head. No conflict and no open change request.
Problem / Motivation
Goal: Each Feature previews switch takes its value from the gateway alone, now that a stable release has shipped the move of those switches to the gateway.
#22913 moved the Settings > Developer > Feature previews switches into gateway config (
dashboard.feature_previews.<name>), with localStorage as a cache. To keep a switch someone had turned on before that change, it added a one-time migration: a browser-only ON is carried up withif_unset, akc-feature-previews-syncedmarker and a per-switchkc-feature-previews-server-backedrecord decide whether a cache-only ON is a value the gateway dropped, and two notices report a refused carry. v0.12.0, the first stable release with #22913, is out, so the migration has done its job.Why it matters
The migration is most of
usePreviewFlagServerSync.ts: two extra browser keys, carry counters and two notice strings in every catalog, all to answer "is this ON from before the gateway kept it?". While it stays, the gateway is not the only source of truth: a stale browser ON can still be written back to the gateway, and every later change to the sync has to reason about the unsynced case.Not a goal
if_unsetsupport on the config PATCH: the sidebar session sort still uses it.mc-preview-instance-sessions) and itsuseRemoteCrewChatfallback; this PR only stops carrying it up.What changed (motivation → approach → change)
if_unsetkc-feature-previews-syncedandkc-feature-previews-server-backeddecide whether a cache-only ON was droppedwebsite/src/i18n; the refused-toggle notices are unchangedreconcilePreviewFlagsis now one rule: each switch takes the gateway's value, off when the gateway holds none, unless a toggle of that switch is still in flight. A config read that fails still leaves the cache alone. The comments inutils/previewFlags.tsand the Preview-gated surfaces paragraph oflearn-cron-dashboard.mdsay the same.Backwards compatibility
Breaking: a Feature previews switch that is on only in a browser that has not opened the dashboard since before #22913 now reads as off instead of being saved to the gateway; turning it on again in Settings > Developer > Feature previews saves it. Writer sweep on origin/main @ bab2387: the only writers of
kc-feature-previews-syncedandkc-feature-previews-server-backedweremarkSyncedandmarkServerBackedinwebsite/src/hooks/usePreviewFlagServerSync.ts, both removed; the only writer ofPreviewFlagWriteError.carrywas the carry path ofpersistin the same file, removed;git grepfinds no other writer inwebsite/src,src/kiro_crewor the app bundles.Reader: website/src/pages/settings/FeaturePreviewsSection.tsx:FeaturePreviewsSection -- chat -- featurePreviewsSectionNotice.test.tsx (the refused-toggle notices still name the switch)
Reader: website/src/hooks/usePreviewFlag.ts:useRemoteCrewChat -- chat -- previewFlagServerSync.test.tsx "clears the retired Remote crew sessions key on load, even when the config cannot be read"
Reader: website/src/shell/appShell.tsx:usePreviewFlagServerSync -- chat -- previewFlagServerSync.test.tsx (the hook's signature is unchanged)
Tests
previewFlagServerSync.test.tsxdrops the carry-up cases and keeps reconcile, in-flight protection and refused-toggle revert. New: a switch on only in this browser reads as off with no write, and both retired keys are cleared on load. The retired-key test now pins that the key is cleared on load with no write, even when the config read fails. These and the toggle test (no third PATCH argument) fail against the old hook.featurePreviewsSectionNotice.test.tsxdrops the two carry-notice cases and no longer seeds the synced marker.tsc -p tsconfig.app.json, eslint,i18n:check,i18n:pseudo(no diff), semgrep, docs-lint.Manual verification
N/A — unit coverage sufficient: the change removes a migration path, and the tests drive the sync hook and the section through the same query client and storage the app shell uses.
Screenshots / video
Why no screenshot: the only rendered change is that the two refused-carry notices can no longer appear; every state that still renders is unchanged.
Related Issues
Closes #24963
Checklist
feat|fix|docs|style|refactor|perf|test|chore|ci|build|revert: ...)Contribution License Agreement
Contributions are accepted under the project's Apache-2.0 license; see the Licensing section of CONTRIBUTING.md.