perf(messages): cut main-thread work while a busy channel is open - #2729

…leanup

* origin/main:
  test(mentions): serve the mentions fixture as a build so cold loads fit (#2776)
  test(browser): let a late publication land after its page closed (#2565)
  test(relay): wait for the verified owner before a guest edits an agent (#2413)
  test: wait for the companion default width measurement (#2750)
  test(browser): preserve conversation scroll while revealing actions (#2715)
  test(browser): wait for link hit-testing after thread close (#2792)
  perf(messages): skip micromark for plain prose bodies (#2780)
  perf(inbox): load the emoji name table on first use (#2774)
  perf(messages): cut main-thread work while a busy channel is open (#2729)
  Refine agent Attention cards, watch details, and status controls (#2717)
  fix(bestie): remove a Bestie from its agent card on any device (#2720)
  feat(agents): preview Builderlab agent card links with Add to Buzz (#2719)

…h-up dwell (#2837)

## Problem

`ThreadPanel.test.tsx` › "observes a visible reply before its root
resolves, then enables thread catch-up" fails deterministically on main
since fca0629 (#2729).

Bisect: passes at `fca0629fb^`, fails at `fca0629fb` and at the current
main tip.

## Cause

#2729 reads `latestMessageId` through `useEffectEvent` instead of
restarting the observation effect. When the root resolves, its row
measures in the next frame. That list mutation triggers `reflow`, which
cancels and re-arms the 300 ms dwell one frame later. Catch-up still
fires exactly once, one frame later than the test expected.

## Change

- The test advances the extra measurement frame before asserting the 299
ms / 300 ms dwell boundary. No product code changes.

## Checks

- `vitest run src/features/messages/ThreadPanel.test.tsx`: 91/91 pass, 3
runs.
- `biome check` clean.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Signed-off-by: Kalvin Chau <[email protected]>
Co-authored-by: Claude Opus 5.5 <[email protected]>

Co-authored-by: Eva <011987e296fd5006292d2f930b574be47c7801048d1983c46c425d3c95f0cffd@buzz.block.builderlab.xyz>
Co-authored-by: Tyler Longwell <[email protected]>
Signed-off-by: Tyler Longwell <[email protected]>

* origin/main:
  test(messages): wait for the row measurement frame before thread catch-up dwell (#2837)
  Keep agent coordination history readable after work finishes (#2454)
  test(mentions): serve the mentions fixture as a build so cold loads fit (#2776)
  test(browser): let a late publication land after its page closed (#2565)
  test(relay): wait for the verified owner before a guest edits an agent (#2413)
  test: wait for the companion default width measurement (#2750)
  test(browser): preserve conversation scroll while revealing actions (#2715)
  test(browser): wait for link hit-testing after thread close (#2792)
  perf(messages): skip micromark for plain prose bodies (#2780)
  perf(inbox): load the emoji name table on first use (#2774)
  perf(messages): cut main-thread work while a busy channel is open (#2729)
  Refine agent Attention cards, watch details, and status controls (#2717)
  fix(bestie): remove a Bestie from its agent card on any device (#2720)
  feat(agents): preview Builderlab agent card links with Add to Buzz (#2719)
  Restore macOS Cmd-W window closing when no tab handles it (#2443)
  Make agent transcripts quieter and link finished replies to activity (#2424)

…ve-enter

* origin/main: (21 commits)
  Make drafts and sent part of Messages (#2728)
  fix(canvas): render every Canvas within the save limit (#2839)
  test(messages): wait for the row measurement frame before thread catch-up dwell (#2837)
  Keep agent coordination history readable after work finishes (#2454)
  test(mentions): serve the mentions fixture as a build so cold loads fit (#2776)
  test(browser): let a late publication land after its page closed (#2565)
  test(relay): wait for the verified owner before a guest edits an agent (#2413)
  test: wait for the companion default width measurement (#2750)
  test(browser): preserve conversation scroll while revealing actions (#2715)
  test(browser): wait for link hit-testing after thread close (#2792)
  perf(messages): skip micromark for plain prose bodies (#2780)
  perf(inbox): load the emoji name table on first use (#2774)
  perf(messages): cut main-thread work while a busy channel is open (#2729)
  Refine agent Attention cards, watch details, and status controls (#2717)
  fix(bestie): remove a Bestie from its agent card on any device (#2720)
  feat(agents): preview Builderlab agent card links with Add to Buzz (#2719)
  Restore macOS Cmd-W window closing when no tab handles it (#2443)
  Make agent transcripts quieter and link finished replies to activity (#2424)
  fix(channels): wrap unread conversation stepping around the sidebar (#2682)
  fix(navigation): let bare buzz:// open Buzz without a broken-link notice (#2718)
  ...

Co-authored-by: BuzzPRReviewAssistant1 <a581a7e8115222acdc32ffb14bfce55d48c24478d98e155f8eba126b69b83c7b@buzz.block.builderlab.xyz>

* origin/main: (71 commits)
  test(activity): scope preview assertions to the request row (#2843)
  perf(search): fold people names once across palette keystrokes (#2570)
  fix(github): keep PR watches running when localStorage is full (#2699)
  Make drafts and sent part of Messages (#2728)
  fix(canvas): render every Canvas within the save limit (#2839)
  test(messages): wait for the row measurement frame before thread catch-up dwell (#2837)
  Keep agent coordination history readable after work finishes (#2454)
  test(mentions): serve the mentions fixture as a build so cold loads fit (#2776)
  test(browser): let a late publication land after its page closed (#2565)
  test(relay): wait for the verified owner before a guest edits an agent (#2413)
  test: wait for the companion default width measurement (#2750)
  test(browser): preserve conversation scroll while revealing actions (#2715)
  test(browser): wait for link hit-testing after thread close (#2792)
  perf(messages): skip micromark for plain prose bodies (#2780)
  perf(inbox): load the emoji name table on first use (#2774)
  perf(messages): cut main-thread work while a busy channel is open (#2729)
  Refine agent Attention cards, watch details, and status controls (#2717)
  fix(bestie): remove a Bestie from its agent card on any device (#2720)
  feat(agents): preview Builderlab agent card links with Add to Buzz (#2719)
  Restore macOS Cmd-W window closing when no tab handles it (#2443)
  ...

Signed-off-by: Ravneet Arora <[email protected]>

…ime-hint

* upstream/main: (26 commits)
  Make drafts and sent part of Messages (block#2728)
  fix(canvas): render every Canvas within the save limit (block#2839)
  test(messages): wait for the row measurement frame before thread catch-up dwell (block#2837)
  Keep agent coordination history readable after work finishes (block#2454)
  test(mentions): serve the mentions fixture as a build so cold loads fit (block#2776)
  test(browser): let a late publication land after its page closed (block#2565)
  test(relay): wait for the verified owner before a guest edits an agent (block#2413)
  test: wait for the companion default width measurement (block#2750)
  test(browser): preserve conversation scroll while revealing actions (block#2715)
  test(browser): wait for link hit-testing after thread close (block#2792)
  perf(messages): skip micromark for plain prose bodies (block#2780)
  perf(inbox): load the emoji name table on first use (block#2774)
  perf(messages): cut main-thread work while a busy channel is open (block#2729)
  Refine agent Attention cards, watch details, and status controls (block#2717)
  fix(bestie): remove a Bestie from its agent card on any device (block#2720)
  feat(agents): preview Builderlab agent card links with Add to Buzz (block#2719)
  Restore macOS Cmd-W window closing when no tab handles it (block#2443)
  Make agent transcripts quieter and link finished replies to activity (block#2424)
  fix(channels): wrap unread conversation stepping around the sidebar (block#2682)
  fix(navigation): let bare buzz:// open Buzz without a broken-link notice (block#2718)
  ...

Signed-off-by: Cheng Peng <[email protected]>
Co-authored-by: Cursor <[email protected]>

## Failing test

`src/app/navigation.test.tsx` "keeps a return path for addressed plugin
settings and retires it on Close" (added by #2491) fails intermittently
in the Vitest flake-check lane:

```
Error: Test timed out in 5000ms.
```

It is one long journey on a full `App` mount with the default 5000 ms
timeout: open the Plugins list, open Pair mobile, Back, Forward, open
Appearance, open Pair mobile, Close, Back, Back.

## Why this is a timeout flake, not a race

The test is slow all the time, and every failure is the same timeout. No
assertion has failed.

- **Main:** 25 green `JavaScript` shard runs since #2491 took 2148–4414
ms (median 3921 ms). That leaves under 1.2 s of headroom.
- **Flake-check failures:** each one was a cold iteration 1 that ran
past 5000 ms. The same test then passed iterations 2–5 at its usual
speed, with no product or test change between them.

| Run | Iter 1 | Iters 2–5 |
|---|---|---|
| #2718
[42217828904](https://github.com/block/buzz-app/actions/runs/42217828904),
Flake check 1/2 | **× 5045 ms** | 4061–4133 ms |
| #2718, Flake check 2/2 | 3796 ms | 3377–3581 ms |
| #2729
[42194107421](https://github.com/block/buzz-app/actions/runs/42194107421)
attempt 1 | **× 5055 ms** | 3973–4399 ms |
| #2729 attempt 2 | **× 5224 ms** | 4016–4423 ms |

- **Neither PR touches this test's path.** #2718 changes deep-link
ingress and #2729 changes messages and relay code. The flake check
selected this file only through imports (`navigation.test.tsx` →
`services.ts` → `deep-links.ts`). #2729's attempt 3 passed.
- **Local near-miss:** in the local baseline below, one iteration on
unmodified `main` took 4920 ms.

## What this changes

The test is split where its title already splits. Each new test mounts
its own `App` with the file's usual per-test setup. The 5000 ms timeout
is unchanged.

1. **"keeps a return path for addressed plugin settings through Back and
Forward":** open the list, open Pair mobile, Back, Forward. These are
the original's first-half assertions, verbatim.
2. **"retires every visit to addressed plugin settings on Close":** open
the list, open Pair mobile, open Appearance, open Pair mobile, Close,
Back, Back. These are the original's second-half assertions, verbatim.
- The test keeps one earlier Pair mobile visit behind Appearance.
`dismissPluginDetails` retires *every* matching visit, and the final
Back landing on the list is what proves it.
- Without the earlier visit, that last assertion is vacuous. With Close
mutated to retire only the current visit
(`navigation.dismiss([navigation.snapshot().entry.id], settings)`):

| Version | Result with the mutated Close |
|---|---|
| Original test | fails |
| Split without the earlier visit | **passes** (misses the regression) |
| This split | fails |

Every assertion in the original test is kept in one of the two tests.
Production code is unchanged.

## Verification

All runs were local, on macOS (16 cores). The flake-check runs happened
while the machine was under heavy unrelated load (load average 35–40).

- `tsc --noEmit` and `biome check --error-on-warnings
src/app/navigation.test.tsx` pass.
- Each new test passes run alone (`-t`), and the whole file passes under
`--sequence.shuffle` with seeds 1–3. Seed 2 runs the Close test first.

**Per-test timings (ms).** The command was `BUZZ_TEST_WORKERS=2
./bin/node scripts/flake-check.mjs vitest --iterations=5 <files>`. The
file list is the same 15-file selection as #2718's failing lane.

| Test | 1 | 2 | 3 | 4 | 5 |
|---|---|---|---|---|---|
| Before: original (on `3bcf1c4bb`) | 2407 | 1994 | 2487 | 2037 | 4920 |
| After: Back and Forward | 1399 | 2071 | 1538 | 906 | 1384 |
| After: Close | 1951 | 2966 | 2131 | 1496 | 1801 |

Running `navigation.test.tsx` alone the same way: Back and Forward took
848–1595 ms and Close took 1352–2114 ms, with 0 of 5 iterations failing.

**Headroom.** The local worst case drops from 4920 to 2966 ms. The Close
half is the longer one, at about 0.75–0.8× the original's local time.
Scaled to CI's 3921 ms median, it should land near 3 s, about 1.8–2 s
under the timeout instead of about 1.1 s. That is an estimate. This PR's
own flake-check lane will repeat the edited file and give the real CI
numbers.

**Unrelated local failures.** Each 15-file run had one failure in
`src/app/entity-navigation.test.tsx`, which this PR does not touch, and
both happened under the load above:
- Before: "opens Inbox for a malformed launch link when the catalog
settles after the client" failed a `findByRole` wait.
- After: "normalizes old Bestie links … retaining { …(2) } scope" timed
out at 5608 ms.

No browser journeys were added or removed.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Signed-off-by: Matt Toohey <[email protected]>
Co-authored-by: Claude Opus 5.5 <[email protected]>