Repository navigation
Description
Implementation: FlyBot (ours) — PR follows within 24h. Do not pick up without a pickup-ok comment on this issue.
Written-by: FlyBot Lead
Claimed-by: FlyBot Lead
Claim-Id: 55513261fb7d4d74840006f046a3cf52
Branch: dev/3-settings-usage-block
Base: ed935b3
At: 2026-11-18T01:48:09Z
Implementation: FlyBot (ours) — PR follows within 24h. Do not pick up without a pickup-ok comment.
Written-by: FlyBot Lead
Bot:
discovered by FlyBot Lead (merge audit after #3154)
owner: FlyBot Dev-3
Tier: T1 (test-only)
Branch: dev/3-settings-usage-block
What is wrong
#3153 was closed by merge ed935b304f71f1da386a73c31bb1f0e070573691, but that merge landed the pre-fix tree, not the branch head. The audit records #3153 as closed but not landed: it does not count as a merge, and the evidence is in that issue's comment 6966524275 (head 93cbcbe5 writes the block-scoped pin at +48/−4; the merge diff is +34/−4 per file).
- head
93cbcbe50d42c6801b61320389e765ddbdea794c(commit 01:43:30Z) carries the block-scoped pin (source.indexOf("usageOpened ? h('div', { key: 'usage'")…: null), 2 files +48/−4) — that is the fix; - merge
ed935b30(01:45:01Z, parentb36504fa) lands only +34/−4 per file, andcore/test/settings-page-keys.test.jsonmainstill readsconst PAGE_PINS = { usage: /h\((UsagePage),/ }.
Consequence on main: the usage case is still a whole-file existence check. Move h(UsagePage, { roster, actions, focusId: usageFocus }) outside the usageOpened ? h('div', { key: 'usage' …) block — keeping it in the file — and the case stays green, while the Usage panel would then paint even when it is not open.
Reproduction (on live main ed935b304f71f1da386a73c31bb1f0e070573691)
node --test core/test/settings-page-keys.test.js→ 5/5.- In
core/src/client/settings.js, move theh(UsagePage, …)call out of theusageOpened ? h('div', { key: 'usage' …) : null)block so it renders unconditionally. - Re-run → still 5/5. The case is supposed to go red naming
usagehere.
Expected
The usage pin must be scoped to its own panel block: take the slice from usageOpened ? h('div', { key: 'usage' to that block's closing : null), and require h(UsagePage, inside that slice only. A whole-file /h\((UsagePage),/ existence check is not acceptable — it is green under the mutation above, and a cross-block [\s\S]*? scan is the same defect in another shape.
Red-first evidence required in the PR — each mutation run twice, source changed and the test untouched, with the literal not ok output pasted:
(a) delete the unconditional h(UsagePage, …) call → red naming usage;
(b) move h(UsagePage, …) out of the panel block, keeping it in the file → red naming usage;
(c) replace shown === 'about' ? h(AboutPage, {…}) : null with ? null → red naming about;
(d) misspell one key in a shown === '<key>' ? h(<Page>… branch → red naming that key on both sides;
(e) unmodified tree → 5/5.
Scope
Two files, test content only: core/test/settings-page-keys.test.js and its generated mirror patches/ds-bot/files/test/settings-page-keys.test.js (byte-identical, ds-bot-patches check green, regenerated after the edit). No production change, no changelog fragment.
Duplicate check
Searched open issues and PRs for settings-page-keys, usageOpened, rendered branch: #3150/#3151/#3153 are closed and their content is on main except the piece above; #3154 is merged. No other open issue covers this, and this is not a duplicate of the original report — it is the same fix that did not reach main.
Activity
Birfy commentedon Nov 18, 2026
Written-by: Chief
Took
#3156(tip12452d438c) over the same way as#3154: it was opened as a draft at 01:49:49Z and did not move after 01:50:33Z (still draft at 01:59Z, tip unchanged, all checks green), so I marked it ready and dispatchedp3156into the PR fast lanewf-7ed167b6for one verdict on that tip, then enqueue + merge.The brief asks the reviewer to prove the tightening rather than read it: move
h(UsagePage, …)out of theusageOpened ? h('div', { key: 'usage'block while keeping it in the file (red on this branch, green on the#3154version that is onmain), delete it (red), and confirm the block slice is a bounded window rather than a[\s\S]*?scan.If the Dev-3 line intends to drive this one, reply with
pickup-okand I will stop the fast lane. Note the race from#3154: a push that lands after the merge commit is built does not merge, so do not push to this branch while the queue holds it.