perf(client-runtime): speed up message sync on desktop and mobile

[Bil0000]

Message catch-up applied each event to the UI separately. A 500-delta replay caused 500 message-list updates, keeping desktop and mobile busy while showing "Syncing messages...".

Apply each received batch in order with the existing reducer, then publish and save its final state and cursor once. Single events keep the existing path. Snapshots, reverts, deletes, and pending older-page merges retain their per-event handling. The shared client runtime covers web, desktop, and mobile without changing the protocol, retry policy, or network buffering.

Measured with 500 text deltas through the real thread state machine and message atom, alternating the baseline (7bd7f99e6c) and this branch for nine runs per batch size and discarding the first pair:

Received batch size	Baseline median	PR median	Message-list updates
16	72.02 ms	8.07 ms	500 → 32
500	50.89 ms	1.36 ms	500 → 1

These measure client replay processing, not download time or device battery use. Slow links still determine when bytes arrive; no timer waits for a batch to fill.

Validation: 106 focused tests passed across thread sync, atoms, pagination, and the reducer. The regression checks batch sizes 1/16/500, complete ordered text, duplicate delivery, the reconnect cursor, and older-page merging before later deltas. Client-runtime, web, and mobile type checks, targeted lint/format, and the production web build passed.

A real web-client check at 1440 px and 390 px replayed 500 synthetic text deltas, verified the complete text and cleared sync banner, and reported no browser errors. Native mobile is covered by shared-runtime tests and its type check; it was not run on a device.

Model: GPT-6-Astra. Harness: Codex.

[macroscopeapp]

Approvability

Verdict: Approved at bcda443

Macroscope's review found this PR approvable — This is a narrowly scoped client-runtime performance improvement that coalesces existing thread synchronization updates while preserving event ordering and exceptional handling. The accompanying tests cover the batching behavior and pagination continuity, with no product-default, schema, security, or deployment changes.

^{You can add or adjust custom eligibility rules. Learn more.}

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Advanced

Run ID: a60ba51a-bce4-435c-aac3-882e144999ca

📥 Commits

Reviewing files that changed from the base of the PR and between e1c94f7 and bcda443.

📒 Files selected for processing (3)

• packages/client-runtime/src/state/threads-atoms.test.ts
• packages/client-runtime/src/state/threads-pagination.test.ts
• packages/client-runtime/src/state/threads.ts

Included review availability: Your plan provides up to 10 included reviews per hour; 2 remain after this review.

---

📝 Walkthrough

Walkthrough

Thread stream consumption now supports batched items. Ordinary batches produce one state update, while complex batches retain per-item handling. Tests cover publication counts, cursor advancement, remounts, and live updates after page merges.

Changes

Thread batch processing

Layer / File(s)	Summary
Batched state reduction packages/client-runtime/src/state/threads.ts	Thread processing reduces ordinary event batches into one state update and preserves per-item handling for complex batches.
Stream integration and regression coverage packages/client-runtime/src/state/threads.ts, packages/client-runtime/src/state/threads-atoms.test.ts, packages/client-runtime/src/state/threads-pagination.test.ts	Stream consumption selects single-item or batched processing. Tests verify batch publication counts, cursor advancement, remount behavior, and live deltas after page merges.

Priority: ➖ Normal

Estimated code review effort: 3 (Moderate) | ~25 minutes

Change: Refactor

Suggested reviewers: juliusmarminge

Sequence Diagram(s)

sequenceDiagram
  participant ThreadStream
  participant applyItems
  participant ThreadState
  ThreadStream->>applyItems: Deliver a batch of thread items
  applyItems->>ThreadState: Apply one reduced state update
  ThreadState-->>ThreadStream: Advance sequence and synchronization state

Loading

Merge Risk: ⚪ Minimal · up to 8281a

The batched stream path preserves the complex-event fallback and is covered for publication, cursor, remount, and page-merge behavior. No merge-blocking risk is identified.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check. Docstring coverage is scoped to functions touched by this diff. Analyzed 0 functions across 3…
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Title check	✅ Passed	The title clearly identifies the primary change: improving message synchronization performance across client platforms.
Description check	✅ Passed	The description explains what changed, why it changed, performance results, scope, validation, and UI verification. It does not use the template headings or checklist, but it provides the required inf…

✨ Finishing Touches 💡 1

🛠️ Fix failing CI checks 💡

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

_{Comment @coderabbitai help to get the list of available commands.}