perf(web): speed up terminal snapshots

[t3dotgg]

Busy terminals allocate a WASM grapheme buffer and build several memory views for each cell in a snapshot.

Reuse one growing grapheme buffer per terminal. Read each cell's style through one view and reuse scalar memory views, refreshing them when shared WASM memory grows. The terminal ABI and output are unchanged.

Source benchmark with the real WASM, Node 24.20, a full 120x40 grid, 10 warmups, and 60 samples:

Per snapshot	Before	After
Median time	15.45 ms	6.58 ms
WASM calls	43,057	33,775
WASM scratch allocations after warmup	4,641	0

These are source benchmarks, not browser frame measurements. All 13 baseline snapshot cases matched, including styles, Unicode, selection, memory growth, resize, scroll, and reset.

Verified with 71 focused Ghostty tests, the web typecheck, and targeted lint. New tests check buffer reuse, cleanup, and styles after memory growth. No browser or device testing was used.

Part of the performance audit.

Created with GPT-6 Astra (preview) in Codex.

---

Note

Medium Risk
Touches the hot snapshot/render path and WASM memory lifetime; behavior is intended to be identical but regressions would affect terminal display and selection styling.

Overview
Ghostty terminal snapshots are faster by cutting per-cell WASM scratch allocations and redundant memory views, without changing the snapshot API or cell output.

Each GhosttyTerminalCore now keeps a reusable grapheme buffer that grows when needed and is freed on dispose(), instead of allocating and freeing a codepoint buffer for every non-empty cell. ghosttyCellText adds a single-codepoint fast path for the common case.

The runtime caches a DataView over WASM memory and refreshes it when memory.grow replaces the buffer; scalar readField / setField / readPointer use that view. During row reads, style flags are decoded from a DataView after grapheme and wide-cell ABI work so styles, selection, and wide-cell spacers stay correct when shared memory grows mid-snapshot.

New WASM-backed tests cover buffer reuse, disposal, and correctness after artificial memory growth.

^{Reviewed by Cursor Bugbot for commit e0e5dc0. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Speed up terminal snapshots in GhosttyTerminalCore

• GhosttyTerminalCore row snapshots reuse a growable WASM buffer for grapheme extraction and read style/selection data from a refreshed DataView after ABI calls.
• GhosttyRuntime caches a full-memory DataView and updates it automatically when WASM memory grows, fixing detached views after memory growth.
• GhosttyTerminalCore.dispose releases the persistent grapheme buffer.
• Adds tests for snapshot rendering after memory growth, wide and astral characters, and grapheme buffer disposal.

^{Macroscope summarized e0e5dc0.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	13.5 KiB	13.7 KiB	+235 B (+1.7%)	15.1 KiB	✅
Codex	Thread snapshot wire	7.0 KiB	7.0 KiB	−11 B (−0.2%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.5 KiB	6.7 KiB	+246 B (+3.7%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	57.0 KiB	58.5 KiB	+1.5 KiB (+2.6%)	66.4 KiB	✅
Codex	Live turn messages	8	10	+2 (+25.0%)	21	✅
Claude	Total thread wire	13.7 KiB	13.8 KiB	+3 B (+0.0%)	15.1 KiB	✅
Claude	Thread snapshot wire	7.0 KiB	7.0 KiB	0 B (0.0%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.7 KiB	6.7 KiB	+3 B (+0.0%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	59.3 KiB	59.3 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	10	10	0 (0.0%)	21	✅

Baseline: caa8a0d · PR result: e0e5dc0 · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 113.8 KiB
• Claude decoded thread snapshot: 114.5 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.

[macroscopeapp]

Approvability

Verdict: Approved at e0e5dc0

Macroscope's review found this PR approvable — This is a contained terminal-rendering performance optimization that reuses WASM buffers and refreshes memory views safely across growth, while preserving the existing snapshot surface. Focused tests cover rendering state, Unicode, selection, memory growth, reuse, and disposal.

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