🤖 perf: defer workspace-open git/PR probes until the chat replay settles

[ThomasK33]

Summary

Opening a workspace no longer starts its git status, git fetch, gh pr view and gh stack view probes while the chat history replay is still running. The probes now start when the chat reaches caught-up. On cold opens, the first transcript row appears ~70-95 ms sooner, and the server replay is 2-3x faster.

Refs #4662

Background

Opening a workspace mounts the footer, branch selector and PR links. Their store subscriptions queued an immediate refresh, which sent four workspace.executeBash calls ~5 ms before the onChat replay began. Each child-process spawn blocks the Electron main process for ~8-16 ms, so the replay's history read ran on a stalled event loop (found while diagnosing #4624; #4670 fixed the untrusted-project part of that measurement).

Implementation

• WorkspaceStore.isWorkspaceChatReplayPending(id) is true from the moment a workspace becomes the active onChat subscription until one of these happens:
  • its caught-up arrives (complete or failed replay);
  • an attempt ends without caught-up (transport error or the 10 s stall watchdog). This is the fallback, so the probes never wait forever on retries;
  • the workspace is deactivated or removed.
    Each of these paths notifies subscribeKey listeners. No timers are involved.
• GitStatusStore and PRStatusStore accept that gate via setChatReplayGate, wired once in AppLoader.
  • Their refresh loops skip a workspace while its replay is pending. This covers every trigger: subscribe, sync, focus and invalidate.
  • For each skipped workspace, one watcher (deferWhileChatReplayPending) calls requestImmediate() when the gate opens.
  • Other workspaces refresh as before.
• Cached git status and the persisted PR badge stay visible while a refresh is deferred.
• Added delay: probes now start at caught-up, ~85-115 ms after the store switch on cold open (median) and ~70-180 ms on switch-back. Switch-back already shows the cached status.

Validation

perf.chatSwitch (--workers 1), 4 alternating base/after runs on the same host, 12 switches per column per build. Medians, in ms:

metric	cold-open-small	cold-open-large	switch-back-small	switch-back-large
renderer.firstRowMs (store switch)	246.9 → 153.1	286.5 → 216.2	25.9 → 28.4	27.1 → 25.3
renderer.caughtUpMs	183.8 → 86.3	173.7 → 114.3	75.3 → 67.3	220.6 → 182.0
caught-up → first row (difference of medians)	63.1 → 66.8	112.8 → 101.9	–	–
dom.firstRowFromClickMs	269.2 → 173.1	291.3 → 223.0	97.3 → 96.8	124.7 → 122.2
dom.longestTaskMs	69 → 75	104 → 107	98.5 → 98	127 → 129.5
server.totalMs	110.8 → 38.0	135.0 → 81.4	39.1 → 9.6	88.4 → 37.4
server.historyReadMs	87.6 → 28.1	94.2 → 50.1	26.7 → 3.1	41.2 → 24.8

The renderer's longest task did not change. In these baseline runs the probe results did not land during the paint (base longest task 69/104 ms, not the 146 ms seen in #4624), so this PR does not claim a longest-task improvement. The numbers were measured at the first commit; the later commits are a behavior-preserving refactor and a test-only rename.

Remote dogfood UAT (Coder Agents, browser mode, real git repo with a GitHub PR) passed:

• On the switch path, the four probes started 155-204 ms after caught-up on this branch; on base they started 36-88 ms before it.
• Footer git status and the PR badge still appeared and stayed correct across switch-back, file changes and restart.
• The three issues it found are pre-existing or by design (they reproduce on base).

Risks

Low. This only changes when the passive git/PR refreshes of the active workspace run. They are late by at most the replay duration, or by one failed attempt (bounded by the stall watchdog) if caught-up never arrives. With no gate wired (tests, stories), behavior is unchanged. The untrusted-project no-repo git env check (#4661) is untouched.

---

Generated with xum • Model: anthropic:claude-opus-5-5 • Thinking: high

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-09-26T12:13:48.367008Z	5c9ff89	PR opened
🔒 Security Review	✅ Completed 2026-09-26T12:15:50.338112Z	5c9ff89	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.