list: batch dynamic height measurements

[Connor Peet (connor4312)]

Improves dynamic-height list rendering by batching DOM writes and reads and by
reusing measured rows that enter the final render range.

• Groups height preparation, measurement, and publication into separate phases.
• Promotes compatible measured rows instead of rendering the same element twice. (biggest win)
• Releases unpromoted rows when rendering fails or the final range changes.
• Validates row identity before promotion after reentrant list updates.
• Adds tests for batching, row reuse, error cleanup, and reentrant splices.

Performance

Measured with DevTools traces while repeatedly alternating between the same small
and large chats. The samples form two non-overlapping modes. The baseline capture
contains 9 small-chat and 10 large-chat swaps; the after capture contains 11
small-chat and 12 large-chat swaps.

Overall interaction

Destination	Metric	Before median	After median	Change
Small chat	Click handler	48.3 ms	33.3 ms	-31.1%
Small chat	Main task	51.0 ms	34.8 ms	-31.9%
Small chat	Thread CPU	47.8 ms	33.9 ms	-29.1%
Large chat	Click handler	291.5 ms	199.3 ms	-31.6%
Large chat	Main task	302.3 ms	204.4 ms	-32.4%
Large chat	Thread CPU	292.0 ms	197.2 ms	-32.5%

The interquartile ranges do not overlap:

Destination	Before p25-p75	After p25-p75
Small chat	43.9-54.3 ms	32.2-35.9 ms
Large chat	266.3-329.3 ms	195.3-204.3 ms

Small-chat pipeline

Stage	Before median	After median	Change
setModel	13.2 ms	10.1 ms	-23.7%
onDidChangeItems	7.4 ms	5.3 ms	-28.6%
Refresh	3.6 ms	2.7 ms	-26.4%
Rerender	4.0 ms	1.1 ms	-71.8%
Dynamic-height probing	3.8 ms	0.8 ms	-78.7%
setChildren	2.2 ms	1.5 ms	-32.2%

Large-chat pipeline

Stage	Before median	After median	Change
setModel	232.7 ms	160.1 ms	-31.2%
onDidChangeItems	205.5 ms	140.7 ms	-31.5%
Refresh	202.0 ms	137.7 ms	-31.9%
Rerender	144.8 ms	95.4 ms	-34.1%
Dynamic-height probing	93.3 ms	66.3 ms	-29.0%
Row insertion	86.2 ms	53.0 ms	-38.5%
Chat row rendering	94.3 ms	61.9 ms	-34.3%
Chat content rendering	66.6 ms	43.8 ms	-34.2%
setChildren	103.3 ms	58.6 ms	-43.3%
Scroll to end	105.0 ms	79.0 ms	-24.8%

Rendering and layout

Destination	Metric	Before median	After median	Change
Small chat	Style recalculation count	11	5	-54.5%
Small chat	Style recalculation time	12.4 ms	8.4 ms	-32.5%
Small chat	Layout count	10	4	-60.0%
Small chat	Layout time	1.8 ms	0.9 ms	-48.4%
Large chat	Style recalculation count	19	13	-31.6%
Large chat	Style recalculation time	92.0 ms	68.0 ms	-26.1%
Large chat	Layout count	18	12	-33.3%
Large chat	Layout time	11.7 ms	7.4 ms	-36.7%

Validation

• npm run eslint -- src/vs/base/browser/ui/list/listView.ts src/vs/base/test/browser/ui/list/listView.test.ts
• npm run typecheck-client
• 10 targeted ListView and ChatListWidget tests
• Independent safety reviews with Claude Opus 5 and GPT-5.6 Terra

(Commit message generated by Copilot)

[Copilot]

Pull request overview

Optimizes dynamic-height list rendering by batching measurements and reusing measured rows.

Changes:

• Separates DOM preparation, measurement, and publication phases.
• Promotes compatible measured rows and cleans up retained rows safely.
• Adds batching, cleanup, and reentrancy tests.

Show a summary per file

File	Description
src/vs/base/browser/ui/list/listView.ts	Implements batched measurement and row promotion.
src/vs/base/test/browser/ui/list/listView.test.ts	Adds dynamic-height regression tests.

Review details

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

• Files reviewed: 2/2 changed files
• Comments generated: 2
• Review effort level: Balanced

[vs-code-engineering]

📬 CODENOTIFY

The following users are being notified based on files changed in this PR:

Benjamin Christopher Simmonds (@benibenj)

Matched files:

• src/vs/base/browser/ui/list/listView.ts
• src/vs/base/browser/ui/tree/abstractTree.ts