Terminal links: accent-coloured underline, text colour unchanged #232

New issue

Closed

opened 2026-10-16 07:27:35 +00:00 by kayg · 84 comments

kayg commented 2026-10-16 07:27:35 +00:00

Owner

Copy link

Owner 2026-10-16 ~13:00 IST: agreed. Terminal links (#214) get an accent-coloured underline; the link text keeps the terminal's colour. Proposal from the orchestrator: underline shown on hover (and while the modifier for opening is held), not permanently, so dense output stays quiet. Web-only.

Owner 2026-10-16 ~13:00 IST: agreed. Terminal links (#214) get an accent-coloured underline; the link text keeps the terminal's colour. Proposal from the orchestrator: underline shown on hover (and while the modifier for opening is held), not permanently, so dense output stays quiet. Web-only.

terminal-link-underline-paper.png

20 KiB

perf-origin-dev-40m-request-default-retention.log

697 B

error-context.md

11 KiB

debug-link-underline-ink.png

21 KiB

terminal-link-underline-ink.png

21 KiB

terminal-link-underline-paper.png

20 KiB

terminal-link-underline-ink.png

21 KiB

terminal-link-underline-ink-375.png

20 KiB

terminal-link-underline-paper-375.png

19 KiB

terminal-link-underline-ink-1440.png

65 KiB

terminal-link-underline-paper-1440.png

62 KiB

baseline-transfer-budget.log

6.5 KiB

candidate-transfer-budget.log

6.6 KiB

perf-origin-dev.log

1.6 KiB

perf-origin-dev.log

1.6 KiB

perf-link-underline.log

1.5 KiB

static-size-origin-build.log

6.2 KiB

static-size-origin.log

401 B

verify-iterate.log

210 KiB

config-isolation-rerun.log

1.4 KiB

verify-fast.log

320 KiB

static-size-candidate.log

404 B

terminal-link-underline-ink.png

21 KiB

terminal-link-underline-paper.png

20 KiB

terminal-link-underline-ink-375.png

20 KiB

terminal-link-underline-paper-375.png

19 KiB

terminal-link-underline-ink-1440.png

65 KiB

terminal-link-underline-paper-1440.png

62 KiB

scroll-review.log

1.8 KiB

wide-review.log

3.4 KiB

idle-terminal-canvas-dpr-1.png

9.3 KiB

idle-terminal-canvas-dpr-1.png

9.3 KiB

idle-terminal-canvas-dpr-2.png

40 KiB

idle-terminal-canvas-dpr-3.png

71 KiB

idle-terminal-canvas-dpr-3.png

71 KiB

idle-target.log

2.3 KiB

transfer-baseline.log

6.5 KiB

transfer-target.log

427 B

perf-baseline-measured.txt

1.6 KiB

perf-target-measured.txt

1.6 KiB

scroll-review-before.log

1.8 KiB

wide-review-before.log

3.4 KiB

wide-review-after.log

2.4 KiB

scroll-review-gate.log

1.2 KiB

scroll-check.log

36 B

renderer-build.log

2.1 KiB

renderer-repro.log

2.1 KiB

wide-check.log

36 B

link-unit-final.log

1.2 KiB

terminal-links-final.log

2.2 KiB

terminal-link-underline-ink.png

21 KiB

terminal-link-underline-ink-375.png

20 KiB

terminal-link-underline-paper-375.png

19 KiB

terminal-link-underline-ink-1440.png

65 KiB

terminal-link-underline-paper-1440.png

62 KiB

production-bundle-postcommit.log

16 KiB

perf-origin-dev-scroll-stall.log

1.1 KiB

perf-origin-dev-scroll-no-movement.log

1.1 KiB

perf-origin-dev-final.log

4 KiB

perf-link-underline-startup-diagnostic.log

1,009 B

idle-terminal-canvas-dpr-2.png

40 KiB

verify-fast-final.log

318 KiB

perf-link-underline-final.log

4 KiB

terminal-link-underline-ink.png

21 KiB

terminal-link-underline-paper.png

20 KiB

link-underline-demo.webm

1.8 MiB

review-scroll-before.log

1.8 KiB

review-wide-before.log

3.4 KiB

terminal-link-underline-ink.png

21 KiB

terminal-link-underline-ink-375.png

20 KiB

terminal-link-underline-paper-375.png

19 KiB

terminal-link-underline-ink-1440.png

65 KiB

terminal-link-underline-paper-1440.png

62 KiB

final-link-tooltip-phone.png

38 KiB

final-link-tooltip-desktop.png

83 KiB

final-link-tooltip-short.png

70 KiB

rebase-final-playwright.log

6.1 KiB

rebase-final-unit.log

1.3 KiB

rebase-final-transfer-budget.log

7.8 KiB

dev-transfer-budget.log

8 KiB

static-baseline.log

40 KiB

perf-server-start-failure.log

842 B

perf-origin-dev-rebased-retry.log

2.9 KiB

perf-origin-dev-final-rebased.log

936 B

perf-origin-dev-final-rebased.log

936 B

origin-dev-perf-run.log

4.2 KiB

static-baseline-merge-base.log

7.6 KiB

candidate-perf-run-failed.log

1.3 KiB

candidate-perf-diagnostic.log

2.2 KiB

candidate-perf-run-final.log

4.3 KiB

verify-fast-final.log

318 KiB

terminal-link-underline-paper.png

20 KiB

terminal-link-underline-paper.png

20 KiB

kayg commented 2026-10-16 16:10:31 +00:00

Author

Owner

Copy link

Starting #232 on branch link-underline, based on origin/dev at 41a3a81abceaa4ea7457d84ee05297ca2244efcc. Read all 113 comments on #214 and the #232 brief; the worktree is clean.

Starting #232 on branch `link-underline`, based on `origin/dev` at `41a3a81abceaa4ea7457d84ee05297ca2244efcc`. Read all 113 comments on #214 and the #232 brief; the worktree is clean.

kayg commented 2026-10-16 16:13:20 +00:00

Author

Owner

Copy link

Finding: the existing renderer draws OSC 8 and detected URL underlines in hard-coded #4A90E2 (apps/tracer/vendor/ghostty-web/dist/ghostty-web.js:1941-1947), while packages/ui/src/themes.ts:247 maps the terminal cursor colour to the theme accent. I’ll route hover/modifier underline painting through that existing theme colour and keep cell foreground colours untouched.

Finding: the existing renderer draws OSC 8 and detected URL underlines in hard-coded `#4A90E2` (`apps/tracer/vendor/ghostty-web/dist/ghostty-web.js:1941-1947`), while `packages/ui/src/themes.ts:247` maps the terminal cursor colour to the theme accent. I’ll route hover/modifier underline painting through that existing theme colour and keep cell foreground colours untouched.

kayg commented 2026-10-16 16:25:08 +00:00

Author

Owner

Copy link

Browser finding: the new underline check sees zero theme-accent pixels after hovering the first URL. terminal-links.spec.ts failed at line 183 after 5 seconds; the existing wrapped/OSC 8 click test and touch test passed (2 passed, 1 failed). Reproducer used TRACER_TEST_PORT=4232. Attached Playwright context: error-context.md. I’m separating renderer behavior from the canvas pixel probe before changing the patch.

Browser finding: the new underline check sees zero theme-accent pixels after hovering the first URL. `terminal-links.spec.ts` failed at line 183 after 5 seconds; the existing wrapped/OSC 8 click test and touch test passed (2 passed, 1 failed). Reproducer used `TRACER_TEST_PORT=4232`. Attached Playwright context: [error-context.md](https://git.kayg.org/attachments/c8fddc67-e37f-4248-8c53-2edce8ffeedc). I’m separating renderer behavior from the canvas pixel probe before changing the patch.

kayg commented 2026-10-16 16:31:38 +00:00

Author

Owner

Copy link

The underline is rendered. The false negative was the probe: at 1× scale, accent #7a9cff over background rgb(13,14,16) produced antialiased edge pixels rgb(67,85,135) and rgb(67,85,136) (244 and 250 pixels in the sampled link cells). Renderer cell geometry was 8×18 CSS px and the hover cursor was pointer. Attached diagnostic capture: . The browser assertion now checks the renderer accent against --accent and detects the stroke's accent blend.

The underline is rendered. The false negative was the probe: at 1× scale, accent `#7a9cff` over background `rgb(13,14,16)` produced antialiased edge pixels `rgb(67,85,135)` and `rgb(67,85,136)` (244 and 250 pixels in the sampled link cells). Renderer cell geometry was 8×18 CSS px and the hover cursor was `pointer`. Attached diagnostic capture: . The browser assertion now checks the renderer accent against `--accent` and detects the stroke's accent blend.

kayg commented 2026-10-16 17:19:06 +00:00

Author

Owner

Copy link

Renderer step complete at 5bc724591c. Added theme.cursor row strokes and retained the singular renderer adapter. Build output: updated ghostty-web 5f0c6b7 with Ghostty 5714ed0 (Bun 1.4.2, Zig 0.15.2). SHA256SUMS check passed for JS, WASM and declarations.

Renderer step complete at 5bc724591c9fe4590551ce37d75a0dee4c43fed9. Added theme.cursor row strokes and retained the singular renderer adapter. Build output: `updated ghostty-web 5f0c6b7 with Ghostty 5714ed0 (Bun 1.4.2, Zig 0.15.2)`. SHA256SUMS check passed for JS, WASM and declarations.

kayg commented 2026-10-16 17:20:05 +00:00

Author

Owner

Copy link

Controller integration complete at 258eb33308. Hover paints one focused link; the platform modifier paints visible links and releases them on keyup or blur. Rows and buffer invalidation clear stale ranges. bun run check passed; focused unit suite: 14 tests, 54,610 assertions.

Controller integration complete at 258eb33308a5484f84d6e7c9fa70126f83520ff4. Hover paints one focused link; the platform modifier paints visible links and releases them on keyup or blur. Rows and buffer invalidation clear stale ranges. `bun run check` passed; focused unit suite: 14 tests, 54,610 assertions.

kayg commented 2026-10-16 17:20:17 +00:00

Author

Owner

Copy link

Browser coverage complete at 873f5d9aa6. TRACER_TEST_PORT=4232 bun x playwright test test/terminal-links.spec.ts: 3 passed in 50.4s. The new test checks both themes, hover/modifier show and hide, accent-token equality, unchanged link text, and captures 393x852, 375x667 and 1440x900 frames. Screenshot attachments follow after production capture review.

Browser coverage complete at 873f5d9aa62c0c75a5b2adcb29276e088807302a. `TRACER_TEST_PORT=4232 bun x playwright test test/terminal-links.spec.ts`: 3 passed in 50.4s. The new test checks both themes, hover/modifier show and hide, accent-token equality, unchanged link text, and captures 393x852, 375x667 and 1440x900 frames. Screenshot attachments follow after production capture review.

kayg commented 2026-10-16 17:20:28 +00:00

Author

Owner

Copy link

Vendor documentation step complete at 7a902273d3. Updated the Ghostty Web patch notes with the active accent and batched-text behavior. git diff --check passed.

Vendor documentation step complete at 7a902273d3d072059d81fb338a4156776cf9375c. Updated the Ghostty Web patch notes with the active accent and batched-text behavior. `git diff --check` passed.

kayg commented 2026-10-16 17:29:30 +00:00

Author

Owner

Copy link

Finding: a default non-preview bundle cannot accept the mock product terminal feed: apps/tracer/src/main.ts:353 gates productTestMode on __CLUSTERS_PREVIEW__. The standalone default-production capture therefore connected but did not render the injected URL; this is the intended compile-out path, not a link regression. The passing Playwright captures use the production-optimized bundle with its test bridge enabled by playwright.config.ts.

Finding: a default non-preview bundle cannot accept the mock product terminal feed: `apps/tracer/src/main.ts:353` gates `productTestMode` on `__CLUSTERS_PREVIEW__`. The standalone default-production capture therefore connected but did not render the injected URL; this is the intended compile-out path, not a link regression. The passing Playwright captures use the production-optimized bundle with its test bridge enabled by `playwright.config.ts`.

kayg commented 2026-10-16 17:29:51 +00:00

Author

Owner

Copy link

Visual review: the terminal link underline follows the theme accent, stays absent when hidden, and leaves the terminal text unchanged. Playwright passed 3/3 on unique port 4232. Captures cover both themes at 393x852, 375x667 and 1440x900:

Visual review: the terminal link underline follows the theme accent, stays absent when hidden, and leaves the terminal text unchanged. Playwright passed 3/3 on unique port 4232. Captures cover both themes at 393x852, 375x667 and 1440x900:      

kayg commented 2026-10-16 17:30:48 +00:00

Author

Owner

Copy link

Production bundle comparison against base 41a3a81abc:

• Total JavaScript: 1,305,476 -> 1,307,178 bytes identity (+1,702); gzip 427,151 -> 427,892 (+741); brotli 362,675 -> 363,572 (+897).
• Critical gzip/brotli delta: +352/+536 B; deferred: +29/+11 B; on-demand class total: +360/+350 B.
  Both transfer-budget runs passed. Receipts: base, candidate.

Production bundle comparison against base 41a3a81abceaa4ea7457d84ee05297ca2244efcc: - Total JavaScript: 1,305,476 -> 1,307,178 bytes identity (+1,702); gzip 427,151 -> 427,892 (+741); brotli 362,675 -> 363,572 (+897). - Critical gzip/brotli delta: +352/+536 B; deferred: +29/+11 B; on-demand class total: +360/+350 B. Both transfer-budget runs passed. Receipts: [base](https://git.kayg.org/attachments/1e03b23f-b629-47c4-b83a-fac95d64149e), [candidate](https://git.kayg.org/attachments/b82b95ce-f34d-4392-b0db-cae13b7717d0).

kayg commented 2026-10-16 17:43:32 +00:00

Author

Owner

Copy link

Perf setup finding: the first base run sent all 10,000 URL rows (350,013 bytes) but the default mock terminal retained only 1,119 scrollback rows, so it aborted before recording any timing. The harness now requests the existing large-history test cap (scrollback=40000000) and will verify that all 10,000 rows are retained before timing. No result from the aborted run is used.

Perf setup finding: the first base run sent all 10,000 URL rows (350,013 bytes) but the default mock terminal retained only 1,119 scrollback rows, so it aborted before recording any timing. The harness now requests the existing large-history test cap (`scrollback=40000000`) and will verify that all 10,000 rows are retained before timing. No result from the aborted run is used.

kayg commented 2026-10-16 17:49:26 +00:00

Author

Owner

Copy link

Follow-up: the proto=product test canvas still retained 1,119 rows with the explicit 40,000,000 scrollback cap (configured=10000), so that run also stopped before timing. The benchmark now uses the plain mock terminal mode used by the existing 10k-history renderer checks; it exercises the same TerminalClient, link controller and renderer while isolating terminal scrolling. Candidate and base use the same mode and cap.

Follow-up: the `proto=product` test canvas still retained 1,119 rows with the explicit 40,000,000 scrollback cap (`configured=10000`), so that run also stopped before timing. The benchmark now uses the plain mock terminal mode used by the existing 10k-history renderer checks; it exercises the same TerminalClient, link controller and renderer while isolating terminal scrolling. Candidate and base use the same mode and cap.

kayg commented 2026-10-16 17:52:21 +00:00

Author

Owner

Copy link

Perf harness finding: the plain terminal retained 9,957 scrollback rows plus its 44 visible rows for the 10,000-line workload, but the hover timer mixed the requestAnimationFrame timestamp with performance.now() and returned negative deltas (-134 to -71 ms). That run is discarded; both timestamps now use performance.now() before any timing is accepted.

Perf harness finding: the plain terminal retained 9,957 scrollback rows plus its 44 visible rows for the 10,000-line workload, but the hover timer mixed the requestAnimationFrame timestamp with `performance.now()` and returned negative deltas (-134 to -71 ms). That run is discarded; both timestamps now use `performance.now()` before any timing is accepted.

kayg commented 2026-10-16 17:54:38 +00:00

Author

Owner

Copy link

Candidate performance measurement did not start: Playwright requested /root/.cache/ms-playwright/chromium_headless_shell-1248/chrome-headless-shell-linux64/chrome-headless-shell, while the perf VM only had Chromium 1228 cached. No timing was accepted. I am installing the runner-matched browser in the VM user cache and will repeat the candidate run.

Candidate performance measurement did not start: Playwright requested `/root/.cache/ms-playwright/chromium_headless_shell-1248/chrome-headless-shell-linux64/chrome-headless-shell`, while the perf VM only had Chromium 1228 cached. No timing was accepted. I am installing the runner-matched browser in the VM user cache and will repeat the candidate run.

kayg commented 2026-10-16 17:56:41 +00:00

Author

Owner

Copy link

Perf comparison: origin/dev 41a3a81 vs link-underline 7a90227, measured on infraternal-perf-test (x64, 4 CPUs, 8,262,770,688 bytes RAM) under systemd-run --user --scope -p CPUWeight=10000. Each run sent 10,000 URL rows / 350,013 bytes and retained 9,957 rows. Origin load average was 0.08/0.42/0.64 at start and 0.22/0.44/0.64 at end; candidate was 0.51/0.39/0.57 at start and 0.62/0.42/0.58 at end.

Hover-to-canvas-change latency (24 samples): origin mean 1.19 ms, p50 1.10 ms, p95 1.70 ms, max 3.30 ms; candidate mean 0.78 ms, p50 0.70 ms, p95 1.00 ms, max 2.00 ms.

Scroll frame intervals (3 x 2 s): origin per-run means 17.09/16.67/17.86 ms and p95 16.8/16.7/33.3 ms; candidate means 16.81/16.67/16.67 ms and p95 16.8/16.8/16.7 ms. Per-run maximums were 33.4/16.8/33.4 ms origin and 33.3/16.8/16.8 ms candidate. The sample does not show a scrolling regression; the third-run origin p95 was a 33.3 ms outlier. Raw receipts: origin, candidate.

Perf comparison: origin/dev `41a3a81` vs `link-underline` `7a90227`, measured on `infraternal-perf-test` (x64, 4 CPUs, 8,262,770,688 bytes RAM) under `systemd-run --user --scope -p CPUWeight=10000`. Each run sent 10,000 URL rows / 350,013 bytes and retained 9,957 rows. Origin load average was 0.08/0.42/0.64 at start and 0.22/0.44/0.64 at end; candidate was 0.51/0.39/0.57 at start and 0.62/0.42/0.58 at end. Hover-to-canvas-change latency (24 samples): origin mean 1.19 ms, p50 1.10 ms, p95 1.70 ms, max 3.30 ms; candidate mean 0.78 ms, p50 0.70 ms, p95 1.00 ms, max 2.00 ms. Scroll frame intervals (3 x 2 s): origin per-run means 17.09/16.67/17.86 ms and p95 16.8/16.7/33.3 ms; candidate means 16.81/16.67/16.67 ms and p95 16.8/16.8/16.7 ms. Per-run maximums were 33.4/16.8/33.4 ms origin and 33.3/16.8/16.8 ms candidate. The sample does not show a scrolling regression; the third-run origin p95 was a 33.3 ms outlier. Raw receipts: [origin](https://git.kayg.org/attachments/42495e9b-0d34-4428-a79d-4374d0eafc40), [candidate](https://git.kayg.org/attachments/336ea79d-c804-43fb-bfb4-787e79b29e4f).

kayg commented 2026-10-16 17:57:47 +00:00

Author

Owner

Copy link

Static size baseline setup: the direct cargo build failed before compilation because .cargo/musl-cc.sh could not find the pinned Zig at $CARGO_TARGET_DIR/auth-tools/zig/zig. scripts/verify stages this repository-pinned toolchain through scripts/auth-toolchain; I am staging it in the job target and will rerun the same x86_64 musl release recipe.

Static size baseline setup: the direct cargo build failed before compilation because `.cargo/musl-cc.sh` could not find the pinned Zig at `$CARGO_TARGET_DIR/auth-tools/zig/zig`. `scripts/verify` stages this repository-pinned toolchain through `scripts/auth-toolchain`; I am staging it in the job target and will rerun the same x86_64 musl release recipe.

kayg commented 2026-10-16 18:09:05 +00:00

Author

Owner

Copy link

Origin/dev static size baseline completed at 41a3a81abceaa4ea7457d84ee05297ca2244efcc. Using the repository release profile and cargo build --locked --release --package infraternal --target x86_64-unknown-linux-musl, the binary is static PIE, stripped, has no PT_INTERP, and is 21,422,096 bytes (the current size ceiling). Build and binary receipts: build log, binary details.

Origin/dev static size baseline completed at `41a3a81abceaa4ea7457d84ee05297ca2244efcc`. Using the repository release profile and `cargo build --locked --release --package infraternal --target x86_64-unknown-linux-musl`, the binary is static PIE, stripped, has no PT_INTERP, and is 21,422,096 bytes (the current size ceiling). Build and binary receipts: [build log](https://git.kayg.org/attachments/449162a4-9e4d-4317-bfda-0670888187fa), [binary details](https://git.kayg.org/attachments/eec8a3c2-5792-45e8-9e82-798c08479361).

kayg commented 2026-10-16 18:23:44 +00:00

Author

Owner

Copy link

verify iterate finished with Rust test failing one case: infraternal-plugin-ssh integration test configured_identity_paths_are_not_inherited_by_the_command_child panicked at crates/plugins/ssh/tests/config_isolation.rs:67 because the synthetic SSH command child could not start (Failure { kind: Install, message: "The SSH helper could not start." }). The remaining Rust summary rows were PASS, including Clippy and cargo-deny. Full gate receipt: verify-iterate.log. I am rerunning the failing test alone before proceeding.

`verify iterate` finished with Rust `test` failing one case: `infraternal-plugin-ssh` integration test `configured_identity_paths_are_not_inherited_by_the_command_child` panicked at `crates/plugins/ssh/tests/config_isolation.rs:67` because the synthetic SSH command child could not start (`Failure { kind: Install, message: "The SSH helper could not start." }`). The remaining Rust summary rows were PASS, including Clippy and cargo-deny. Full gate receipt: [verify-iterate.log](https://git.kayg.org/attachments/254d1bd6-95b7-428f-9a0d-e5923124cd5d). I am rerunning the failing test alone before proceeding.

kayg commented 2026-10-16 18:27:48 +00:00

Author

Owner

Copy link

The isolated retry of configured_identity_paths_are_not_inherited_by_the_command_child passed (1 passed, 0 failed, 0.06 s; no code changes). The first failure occurred during the full workspace run; I will use the required final scripts/verify fast result to confirm the branch gate. Retry log.

The isolated retry of `configured_identity_paths_are_not_inherited_by_the_command_child` passed (1 passed, 0 failed, 0.06 s; no code changes). The first failure occurred during the full workspace run; I will use the required final `scripts/verify fast` result to confirm the branch gate. [Retry log](https://git.kayg.org/attachments/c7cf7ca5-962c-4f68-86b7-5de7c462f2eb).

kayg commented 2026-10-16 19:11:08 +00:00

Author

Owner

Copy link

Finished at head 7a902273d3d072059d81fb338a4156776cf9375c.

The final scripts/verify fast exited 0. All Rust rows passed; the x86_64 static build stayed at 21,422,096 stripped bytes (origin/dev: 21,422,096; delta: 0). The aarch64 static size check also passed. Full 327,282-byte gate transcript: verify-fast.log. Candidate binary details: static-size-candidate.log.

The production web bundle delta against origin/dev is +1,702 identity bytes, +741 gzip bytes, and +897 brotli bytes; the bundle transfer budget passed. Perf VM comparison and raw receipts are in the prior issue comment: 10,000 URL lines / 9,957 retained rows on the same 4-core x64 machine; hover p95 was 1.7 ms origin and 1.0 ms candidate; three scroll-run p95s were 16.8/16.7/33.3 ms origin and 16.8/16.8/16.7 ms candidate. Theme screenshots for all requested viewport sizes were attached earlier. The first iterate run had one SSH helper startup failure; the isolated retry and final full fast run passed unchanged.

Final scripts/verify fast gate summary, verbatim:

x86_64-unknown-linux-musl: exact 21422096-byte cap PASS; 21422097 bytes rejected PASS; smaller artifact requires ratchet PASS
aarch64-unknown-linux-musl: exact 19594856-byte cap PASS; 19594857 bytes rejected PASS; smaller artifact requires ratchet PASS
unsupported target: rejected PASS
server subcommand: exit 2, versioned JSON unavailable result
agent subcommand: exit 2, versioned JSON unavailable result
(static builds: 1222s)

+---------------------------+--------------+
| Rust gate                 | Result       |
+---------------------------+--------------+
| fmt                       | PASS         |
| VT native fmt             | PASS         |
| CLI plugin compile-out    | PASS         |
| clippy                    | PASS         |
| test                      | PASS         |
| cargo-deny                | PASS         |
| static builds             | PASS         |
+---------------------------+--------------+

dist/ghostty-web.js: OK
dist/ghostty-vt.wasm: OK
dist/index.d.ts: OK

 191 pass
 0 fail
 11387 expect() calls
Ran 191 tests across 31 files. [744.00ms]

 37 pass
 0 fail
 34463 expect() calls
Ran 37 tests across 10 files. [5.67s]

Finished at head `7a902273d3d072059d81fb338a4156776cf9375c`. The final `scripts/verify fast` exited 0. All Rust rows passed; the x86_64 static build stayed at 21,422,096 stripped bytes (origin/dev: 21,422,096; delta: 0). The aarch64 static size check also passed. Full 327,282-byte gate transcript: [verify-fast.log](https://git.kayg.org/attachments/e0a6462f-e76d-46ff-bb98-fac39a8b56be). Candidate binary details: [static-size-candidate.log](https://git.kayg.org/attachments/68286258-9c38-4ada-8148-1a5ad9069bd2). The production web bundle delta against origin/dev is +1,702 identity bytes, +741 gzip bytes, and +897 brotli bytes; the bundle transfer budget passed. Perf VM comparison and raw receipts are in the prior issue comment: 10,000 URL lines / 9,957 retained rows on the same 4-core x64 machine; hover p95 was 1.7 ms origin and 1.0 ms candidate; three scroll-run p95s were 16.8/16.7/33.3 ms origin and 16.8/16.8/16.7 ms candidate. Theme screenshots for all requested viewport sizes were attached earlier. The first `iterate` run had one SSH helper startup failure; the isolated retry and final full `fast` run passed unchanged. Final `scripts/verify fast` gate summary, verbatim: ```text x86_64-unknown-linux-musl: exact 21422096-byte cap PASS; 21422097 bytes rejected PASS; smaller artifact requires ratchet PASS aarch64-unknown-linux-musl: exact 19594856-byte cap PASS; 19594857 bytes rejected PASS; smaller artifact requires ratchet PASS unsupported target: rejected PASS server subcommand: exit 2, versioned JSON unavailable result agent subcommand: exit 2, versioned JSON unavailable result (static builds: 1222s) +---------------------------+--------------+ | Rust gate | Result | +---------------------------+--------------+ | fmt | PASS | | VT native fmt | PASS | | CLI plugin compile-out | PASS | | clippy | PASS | | test | PASS | | cargo-deny | PASS | | static builds | PASS | +---------------------------+--------------+ dist/ghostty-web.js: OK dist/ghostty-vt.wasm: OK dist/index.d.ts: OK 191 pass 0 fail 11387 expect() calls Ran 191 tests across 31 files. [744.00ms] 37 pass 0 fail 34463 expect() calls Ran 37 tests across 10 files. [5.67s] ```

kayg commented 2026-10-16 19:15:02 +00:00

Author

Owner

Copy link

Starting independent review on branch review-link-underline, target head 7a902273d3d072059d81fb338a4156776cf9375c, diff base 41a3a81ab.

Starting independent review on branch `review-link-underline`, target head `7a902273d3d072059d81fb338a4156776cf9375c`, diff base `41a3a81ab`.

kayg commented 2026-10-16 19:27:39 +00:00

Author

Owner

Copy link

Independent browser step at target 7a902273d3d072059d81fb338a4156776cf9375c: TRACER_TEST_PORT=4236 bun x playwright test test/terminal-links.spec.ts passed 3/3 (wrapped/OSC 8 links, underline states, touch). The production-optimized tracer build with its test bridge covered both themes at 393×852, 375×667 and 1440×900; six fresh captures are attached below.

Independent browser step at target `7a902273d3d072059d81fb338a4156776cf9375c`: `TRACER_TEST_PORT=4236 bun x playwright test test/terminal-links.spec.ts` passed 3/3 (wrapped/OSC 8 links, underline states, touch). The production-optimized tracer build with its test bridge covered both themes at 393×852, 375×667 and 1440×900; six fresh captures are attached below.

kayg commented 2026-10-16 20:05:02 +00:00

Author

Owner

Copy link

Exact-head verification completed on 7a902273d3 (review-link-underline). No implementation files changed.

Verbatim gate summary:

| Rust gate                 | Result       |
+---------------------------+--------------+
| fmt                       | PASS         |
| VT native fmt             | PASS         |
| CLI plugin compile-out    | PASS         |
| clippy                    | PASS         |
| test                      | PASS         |
| cargo-deny                | PASS         |
| static builds             | PASS         |

 191 pass
 0 fail
 11387 expect() calls
Ran 191 tests across 31 files. [385.00ms]

JavaScript transfer, critical: identity=962310 gzip=298037/307200 brotli=247919/256000 bytes
JavaScript transfer, deferred: identity=137782 gzip=48117/51200 brotli=42669/45056 bytes
JavaScript transfer, onDemand: identity=207086 gzip=81725/20480 brotli=72889/18432 bytes
JavaScript transfer, onDemand largest chunk chunk-kbh28sfh.js: gzip=15943/20480 brotli=14185/18432 bytes (class total gzip=81725 brotli=72889, not limited)

 37 pass
 0 fail
 34463 expect() calls
Ran 37 tests across 10 files. [3.90s]

Exact-head verification completed on 7a902273d3d072059d81fb338a4156776cf9375c (review-link-underline). No implementation files changed. Verbatim gate summary: ``` | Rust gate | Result | +---------------------------+--------------+ | fmt | PASS | | VT native fmt | PASS | | CLI plugin compile-out | PASS | | clippy | PASS | | test | PASS | | cargo-deny | PASS | | static builds | PASS | 191 pass 0 fail 11387 expect() calls Ran 191 tests across 31 files. [385.00ms] JavaScript transfer, critical: identity=962310 gzip=298037/307200 brotli=247919/256000 bytes JavaScript transfer, deferred: identity=137782 gzip=48117/51200 brotli=42669/45056 bytes JavaScript transfer, onDemand: identity=207086 gzip=81725/20480 brotli=72889/18432 bytes JavaScript transfer, onDemand largest chunk chunk-kbh28sfh.js: gzip=15943/20480 brotli=14185/18432 bytes (class total gzip=81725 brotli=72889, not limited) 37 pass 0 fail 34463 expect() calls Ran 37 tests across 10 files. [3.90s] ```

kayg commented 2026-10-16 20:06:30 +00:00

Author

Owner

Copy link

Finding LUR-1 (P2): revealed underline remains in the old screen-relative row after scrolling from absolute rows 100–101 to 500–501. bun test src/terminal-links.on-demand.test.ts reports 13 pass, 1 fail; the new test clears a revealed underline as soon as scrolling moves it out of the viewport observes the stale paint range {startX: 0, startY: 0, endX: 30, endY: 0} before idle scanning. Evidence: $HOME/.cache/codex-scratch/link-underline-review/evidence/scroll-review.log. Test-only commit on local review-link-underline: c57a334bd073e82af4186c39aaa49f8af7ac7790.

Finding LUR-1 (P2): revealed underline remains in the old screen-relative row after scrolling from absolute rows 100–101 to 500–501. `bun test src/terminal-links.on-demand.test.ts` reports 13 pass, 1 fail; the new test `clears a revealed underline as soon as scrolling moves it out of the viewport` observes the stale paint range `{startX: 0, startY: 0, endX: 30, endY: 0}` before idle scanning. Evidence: `$HOME/.cache/codex-scratch/link-underline-review/evidence/scroll-review.log`. Test-only commit on local `review-link-underline`: `c57a334bd073e82af4186c39aaa49f8af7ac7790`.

kayg commented 2026-10-16 20:06:54 +00:00

Author

Owner

Copy link

Vendor reproducibility check completed for target 7a902273d3.

Verbatim result:

reproduced ghostty-web 5f0c6b7 with Ghostty 5714ed0 (Bun 1.4.2, Zig 0.15.2)

apps/tracer/vendor/ghostty-web/build.sh --check compared rebuilt dist and checksums successfully. Log: $HOME/.cache/codex-scratch/link-underline-review/evidence/vendor-rebuild.log.

Vendor reproducibility check completed for target 7a902273d3d072059d81fb338a4156776cf9375c. Verbatim result: ``` reproduced ghostty-web 5f0c6b7 with Ghostty 5714ed0 (Bun 1.4.2, Zig 0.15.2) ``` `apps/tracer/vendor/ghostty-web/build.sh --check` compared rebuilt dist and checksums successfully. Log: `$HOME/.cache/codex-scratch/link-underline-review/evidence/vendor-rebuild.log`.

kayg commented 2026-10-16 20:10:53 +00:00

Author

Owner

Copy link

Finding LUR-2 (P2): the underline stops after the lead cell of a final CJK wide URL glyph. revealed underline covers both cells of a wide final URL glyph at DPR 1, 2, and 3 fails against the production renderer. At endpoint {x:20,y:3}, lead-cell accent pixels are 7/32/89 at DPR 1/2/3; continuation-cell pixels are 0/0/0 (minimums 4/8/12). Location: apps/tracer/vendor/ghostty-web/link-underline.patch:184-199. Evidence: $HOME/.cache/codex-scratch/link-underline-review/evidence/wide-review.log. Test-only commit on local review-link-underline: 389a967c6c139392fc69aeccd7f834a2b5fad75a.

Finding LUR-2 (P2): the underline stops after the lead cell of a final CJK wide URL glyph. `revealed underline covers both cells of a wide final URL glyph at DPR 1, 2, and 3` fails against the production renderer. At endpoint `{x:20,y:3}`, lead-cell accent pixels are 7/32/89 at DPR 1/2/3; continuation-cell pixels are 0/0/0 (minimums 4/8/12). Location: `apps/tracer/vendor/ghostty-web/link-underline.patch:184-199`. Evidence: `$HOME/.cache/codex-scratch/link-underline-review/evidence/wide-review.log`. Test-only commit on local `review-link-underline`: `389a967c6c139392fc69aeccd7f834a2b5fad75a`.

kayg commented 2026-10-16 20:12:45 +00:00

Author

Owner

Copy link

Attached raw review test output: scroll-review.log, wide-review.log.

Attached raw review test output: [scroll-review.log](https://git.kayg.org/attachments/0b9c74eb-3a55-4a41-9227-a279a947ad59), [wide-review.log](https://git.kayg.org/attachments/42e5695b-7ee2-4ba5-9d09-e7df5fecb9cb).

kayg commented 2026-10-16 20:18:03 +00:00

Author

Owner

Copy link

Production renderer idle-pixel comparison completed between feature base 41a3a81ab and target 7a902273d. The terminal contained a long wrapped URL and CJK/emoji output; the pointer rested on a non-link row with no modifier held.

• DPR 1, 424×756: 0 differing pixels, max channel delta 0.
• DPR 2, 848×1512: 0 differing pixels, max channel delta 0.
• DPR 3, 1272×2268: 0 differing pixels, max channel delta 0.

The baseline and target PNGs and raw run log are attached. The Playwright run passed (one test covering all three DPRs).

Production renderer idle-pixel comparison completed between feature base 41a3a81ab and target 7a902273d. The terminal contained a long wrapped URL and CJK/emoji output; the pointer rested on a non-link row with no modifier held. - DPR 1, 424×756: 0 differing pixels, max channel delta 0. - DPR 2, 848×1512: 0 differing pixels, max channel delta 0. - DPR 3, 1272×2268: 0 differing pixels, max channel delta 0. The baseline and target PNGs and raw run log are attached. The Playwright run passed (one test covering all three DPRs).

kayg commented 2026-10-16 20:18:26 +00:00

Author

Owner

Copy link

Bundle comparison completed on the default non-preview builds of feature base 41a3a81ab and target 7a902273d. Identity delta is +1,702 B. The paired gzip totals differ by +715 B: +364 B critical, +24 B deferred, and +327 B on-demand. This run does not reproduce the report's +741 B gzip exactly; the change spans the critical path and on-demand code. The fast gate passed with critical at 298,037/307,200 B gzip and the largest on-demand chunk at 15,943/20,480 B gzip. Raw baseline and target output are attached.

Bundle comparison completed on the default non-preview builds of feature base 41a3a81ab and target 7a902273d. Identity delta is +1,702 B. The paired gzip totals differ by +715 B: +364 B critical, +24 B deferred, and +327 B on-demand. This run does not reproduce the report's +741 B gzip exactly; the change spans the critical path and on-demand code. The fast gate passed with critical at 298,037/307,200 B gzip and the largest on-demand chunk at 15,943/20,480 B gzip. Raw baseline and target output are attached.

kayg commented 2026-10-16 20:54:08 +00:00

Author

Owner

Copy link

Perf VM timing is temporarily blocked by another run holding /var/lock/perf-bench.lock. The review runner is queued with flock -w 3600; no new timing samples have been collected yet. The candidate harness and paired base runner are ready for the lock release.

Perf VM timing is temporarily blocked by another run holding `/var/lock/perf-bench.lock`. The review runner is queued with `flock -w 3600`; no new timing samples have been collected yet. The candidate harness and paired base runner are ready for the lock release.

kayg commented 2026-10-16 21:42:50 +00:00

Author

Owner

Copy link

Perf VM recheck (4 vCPU, 7.7 GiB reported memory; locked systemd scope with CPUWeight=10000). Base: 200 hover samples p95 2.3 ms; scroll frame p95 33.4/33.4/33.4 ms; load averages 2.66/2.03/1.40 at start and 1.92/1.90/1.38 at end. Candidate: hover p95 0.6 ms; scroll frame p95 16.7/16.7/16.7 ms; load averages 1.18/1.57/1.31 at start and 1.30/1.56/1.31 at end. Raw runs: base, candidate.

This does not reproduce the reported 9,957-row scale: both builds received output sequence 620007 after 10,000 URL lines, configured scrollback 10,000, but retained 471 scrollback + 44 visible rows (515 total) and 43 visible links. The timings are supplemental; the baseline window had higher load. A second baseline attempt failed with a socket hang-up before samples and was excluded; its stored diagnostic line was redacted.

Perf VM recheck (4 vCPU, 7.7 GiB reported memory; locked systemd scope with CPUWeight=10000). Base: 200 hover samples p95 2.3 ms; scroll frame p95 33.4/33.4/33.4 ms; load averages 2.66/2.03/1.40 at start and 1.92/1.90/1.38 at end. Candidate: hover p95 0.6 ms; scroll frame p95 16.7/16.7/16.7 ms; load averages 1.18/1.57/1.31 at start and 1.30/1.56/1.31 at end. Raw runs: [base](https://git.kayg.org/attachments/07a98e7c-cc5e-4a86-b89f-2ba659e58d14), [candidate](https://git.kayg.org/attachments/76e81d87-f7be-450e-bbf4-ed2121df99fd). This does not reproduce the reported 9,957-row scale: both builds received output sequence 620007 after 10,000 URL lines, configured scrollback 10,000, but retained 471 scrollback + 44 visible rows (515 total) and 43 visible links. The timings are supplemental; the baseline window had higher load. A second baseline attempt failed with a socket hang-up before samples and was excluded; its stored diagnostic line was redacted.

kayg commented 2026-10-16 21:43:42 +00:00

Author

Owner

Copy link

Verdict: merge. Counts: P0 0, P1 0, P2 2. The P2 failing review tests are committed on review-link-underline: c57a334bd073e82af4186c39aaa49f8af7ac7790 and 389a967c6c139392fc69aeccd7f834a2b5fad75a. Review branch head: 389a967c6c139392fc69aeccd7f834a2b5fad75a; reviewed job head: 7a902273d3d072059d81fb338a4156776cf9375c. No implementation changes were made.

scripts/verify fast exited 0 on the reviewed job head. Verbatim gate summary:

x86_64-unknown-linux-musl: exact 21422096-byte cap PASS; 21422097 bytes rejected PASS; smaller artifact requires ratchet PASS
aarch64-unknown-linux-musl: exact 19594856-byte cap PASS; 19594857 bytes rejected PASS; smaller artifact requires ratchet PASS
unsupported target: rejected PASS
server subcommand: exit 2, versioned JSON unavailable result
agent subcommand: exit 2, versioned JSON unavailable result
(static builds: 1452s)
 191 pass
 0 fail
 11387 expect() calls
Ran 191 tests across 31 files. [385.00ms]
 37 pass
 0 fail
 34463 expect() calls
Ran 37 tests across 10 files. [3.90s]

Verdict: merge. Counts: P0 0, P1 0, P2 2. The P2 failing review tests are committed on `review-link-underline`: `c57a334bd073e82af4186c39aaa49f8af7ac7790` and `389a967c6c139392fc69aeccd7f834a2b5fad75a`. Review branch head: `389a967c6c139392fc69aeccd7f834a2b5fad75a`; reviewed job head: `7a902273d3d072059d81fb338a4156776cf9375c`. No implementation changes were made. `scripts/verify fast` exited 0 on the reviewed job head. Verbatim gate summary: ```text x86_64-unknown-linux-musl: exact 21422096-byte cap PASS; 21422097 bytes rejected PASS; smaller artifact requires ratchet PASS aarch64-unknown-linux-musl: exact 19594856-byte cap PASS; 19594857 bytes rejected PASS; smaller artifact requires ratchet PASS unsupported target: rejected PASS server subcommand: exit 2, versioned JSON unavailable result agent subcommand: exit 2, versioned JSON unavailable result (static builds: 1452s) 191 pass 0 fail 11387 expect() calls Ran 191 tests across 31 files. [385.00ms] 37 pass 0 fail 34463 expect() calls Ran 37 tests across 10 files. [3.90s] ```

kayg commented 2026-10-16 21:45:37 +00:00

Author

Owner

Copy link

Starting the review continuation on branch link-underline, HEAD 7a902273d3d072059d81fb338a4156776cf9375c, based on origin/dev at 41a3a81abceaa4ea7457d84ee05297ca2244efcc. The worktree is clean. I have read all comments on #232 and #214; next I will reproduce both committed review tests and fix the stale-scroll and wide-glyph cases.

Starting the review continuation on branch `link-underline`, HEAD `7a902273d3d072059d81fb338a4156776cf9375c`, based on `origin/dev` at `41a3a81abceaa4ea7457d84ee05297ca2244efcc`. The worktree is clean. I have read all comments on #232 and #214; next I will reproduce both committed review tests and fix the stale-scroll and wide-glyph cases.

kayg commented 2026-10-16 21:47:21 +00:00

Author

Owner

Copy link

Reproduced LUR-1 before the fix at f64c04ddb: bun test src/terminal-links.on-demand.test.ts reports 13 pass / 1 fail. Scrolling the visible range from rows 100–101 to 500–501 leaves the old screen-relative underline {startX: 0, startY: 0, endX: 30, endY: 0} painted before idle scanning. Receipt: scroll-review-before.log.

Reproduced LUR-1 before the fix at `f64c04ddb`: `bun test src/terminal-links.on-demand.test.ts` reports 13 pass / 1 fail. Scrolling the visible range from rows 100–101 to 500–501 leaves the old screen-relative underline `{startX: 0, startY: 0, endX: 30, endY: 0}` painted before idle scanning. Receipt: [scroll-review-before.log](https://git.kayg.org/attachments/d7cf2846-8588-45bc-8543-50beaf3f9b49).

kayg commented 2026-10-16 21:47:47 +00:00

Author

Owner

Copy link

Reproduced LUR-2 before the fix at 563edbb55: production canvas pixels show the underline only in the lead cell of a final CJK glyph. At DPR 1/2/3, lead-cell accent pixels are 7/32/89 and continuation-cell pixels are 0/0/0 (minimum 4/8/12). TRACER_TEST_PORT=4233 bun x playwright test test/renderer-link-wide-review.spec.ts failed as expected. Receipt: wide-review-before.log.

Reproduced LUR-2 before the fix at `563edbb55`: production canvas pixels show the underline only in the lead cell of a final CJK glyph. At DPR 1/2/3, lead-cell accent pixels are 7/32/89 and continuation-cell pixels are 0/0/0 (minimum 4/8/12). `TRACER_TEST_PORT=4233 bun x playwright test test/renderer-link-wide-review.spec.ts` failed as expected. Receipt: [wide-review-before.log](https://git.kayg.org/attachments/2b34c8d8-6615-4e88-b06c-7c63dd01d13f).

kayg commented 2026-10-16 21:49:29 +00:00

Author

Owner

Copy link

LUR-1 is fixed in 3a1a077480a09ce90b188cbed850d501634ef590 and pushed. Viewport updates now clear revealed screen-relative ranges immediately and let the existing idle scan recompute at the current pointer. Focused link tests: 14 pass, 0 fail, 30 assertions. apps/tracer bun run check passed. Receipts: unit output, TypeScript check.

LUR-1 is fixed in `3a1a077480a09ce90b188cbed850d501634ef590` and pushed. Viewport updates now clear revealed screen-relative ranges immediately and let the existing idle scan recompute at the current pointer. Focused link tests: 14 pass, 0 fail, 30 assertions. `apps/tracer` `bun run check` passed. Receipts: [unit output](https://git.kayg.org/attachments/9221729b-132b-462a-95a8-9c5505060827), [TypeScript check](https://git.kayg.org/attachments/a765e6e0-1e94-4c47-8e54-7e7b6402daf6).

kayg commented 2026-10-16 21:56:32 +00:00

Author

Owner

Copy link

LUR-2 is fixed in ebd62f98dd0493cf7a3f7b5f0cdcc5d705a15873 and pushed. The renderer extends the final underline through the lead cell’s existing wide-cell width. The production pixel check passed at DPR 1/2/3: continuation-cell accent pixels 11/32/92 (minimum 4/8/12); lead-cell counts 7/32/89. The renderer build updated the pinned artifact, and build.sh --check reproduced it. apps/tracer TypeScript check passed; focused link tests: 14 pass; browser link suite: 3 pass, covering hover/modifier show and hide, theme accent equality, unchanged text colour, and touch. Evidence: pixel test, build, reproduction check, browser run. Both themes captured at 393x852, 375x667, and 1440x900; the six screenshots are attached above.

LUR-2 is fixed in `ebd62f98dd0493cf7a3f7b5f0cdcc5d705a15873` and pushed. The renderer extends the final underline through the lead cell’s existing wide-cell width. The production pixel check passed at DPR 1/2/3: continuation-cell accent pixels 11/32/92 (minimum 4/8/12); lead-cell counts 7/32/89. The renderer build updated the pinned artifact, and `build.sh --check` reproduced it. `apps/tracer` TypeScript check passed; focused link tests: 14 pass; browser link suite: 3 pass, covering hover/modifier show and hide, theme accent equality, unchanged text colour, and touch. Evidence: [pixel test](https://git.kayg.org/attachments/ebb86bb7-2fce-49dd-8c73-b3359cbac5c1), [build](https://git.kayg.org/attachments/aacbe929-9eb1-4335-a1fa-79caeb4ac874), [reproduction check](https://git.kayg.org/attachments/a2ff4174-1b79-4eeb-a233-b575bb85a2e5), [browser run](https://git.kayg.org/attachments/1ca8fd2d-9065-4e2e-a165-3f881c7f6532). Both themes captured at 393x852, 375x667, and 1440x900; the six screenshots are attached above.

kayg commented 2026-10-16 22:21:05 +00:00

Author

Owner

Copy link

Perf VM wait: at 2026-10-17T00:20:55+02:00, /var/lock/perf-bench.owner reported pill-flicker started=2026-10-16T22:15:31Z pid=1811191; systemctl --user is-active codex-pill-flicker returned inactive. Load average was 1.07 1.24 1.31. I have not changed the lock and continue waiting with flock.

Perf VM wait: at 2026-10-17T00:20:55+02:00, `/var/lock/perf-bench.owner` reported `pill-flicker started=2026-10-16T22:15:31Z pid=1811191`; `systemctl --user is-active codex-pill-flicker` returned `inactive`. Load average was `1.07 1.24 1.31`. I have not changed the lock and continue waiting with `flock`.

kayg commented 2026-10-16 22:46:26 +00:00

Author

Owner

Copy link

Iterate finding: at ebd62f9, scripts/verify iterate failed crates/plugins/ssh/tests/config_isolation.rs::configured_identity_paths_are_not_inherited_by_the_command_child (test gate reported FAIL (101)); the child helper could not start. The same compiled test rerun alone passed: test configured_identity_paths_are_not_inherited_by_the_command_child ... ok (1 passed, finished in 0.04s). No SSH files are changed. I will rerun the full iterate gate to check whether the failure repeats.

Iterate finding: at `ebd62f9`, `scripts/verify iterate` failed `crates/plugins/ssh/tests/config_isolation.rs::configured_identity_paths_are_not_inherited_by_the_command_child` (test gate reported FAIL (101)); the child helper could not start. The same compiled test rerun alone passed: `test configured_identity_paths_are_not_inherited_by_the_command_child ... ok` (1 passed, finished in 0.04s). No SSH files are changed. I will rerun the full iterate gate to check whether the failure repeats.

kayg commented 2026-10-16 22:54:54 +00:00

Author

Owner

Copy link

Iterate rerun finding at ebd62f9: crates/infraternal-server/src/remote_control_queue_tests.rs::saturated_native_control_queue_does_not_hide_primary_terminal_output failed. Receipt: CONTROL_SATURATION active=true queued=1 native_primary_live=true shadow_output=0; panic: full control FIFO prevented healthy primary terminal delivery. The server test group reported 448 passed, 1 failed, 48 ignored in 283.79s. It is outside the changed files; I will rerun this exact test alone to determine repeatability.

Iterate rerun finding at `ebd62f9`: `crates/infraternal-server/src/remote_control_queue_tests.rs::saturated_native_control_queue_does_not_hide_primary_terminal_output` failed. Receipt: `CONTROL_SATURATION active=true queued=1 native_primary_live=true shadow_output=0`; panic: `full control FIFO prevented healthy primary terminal delivery`. The server test group reported 448 passed, 1 failed, 48 ignored in 283.79s. It is outside the changed files; I will rerun this exact test alone to determine repeatability.

kayg commented 2026-10-16 22:55:14 +00:00

Author

Owner

Copy link

The failed server test passed alone on the same compiled binary: 1 passed; 0 failed; 0 ignored; 496 filtered out; finished in 0.16s. The full iterate run remains failed, so this does not clear that gate. At the isolated rerun the build-host load average was 20.23/16.19/17.83. I will rerun the full iterate gate after shared host load drops.

The failed server test passed alone on the same compiled binary: `1 passed; 0 failed; 0 ignored; 496 filtered out; finished in 0.16s`. The full iterate run remains failed, so this does not clear that gate. At the isolated rerun the build-host load average was 20.23/16.19/17.83. I will rerun the full iterate gate after shared host load drops.

kayg commented 2026-10-16 23:07:30 +00:00

Author

Owner

Copy link

Third iterate finding at ebd62f9: crates/infraternal/tests/agent_queues.rs::saturated_terminal_queues_preserve_every_byte_and_daemon_liveness failed its queue saturation oracle. The fixture received {"error":"terminal unavailable","reason":"resource_unavailable"}; test result was 0 passed, 1 failed in 7.38s. Build-host load was 45.05/31.31/23.12. No infraternal/server sources are changed. I will retry after concurrent jobs clear.

Third iterate finding at `ebd62f9`: `crates/infraternal/tests/agent_queues.rs::saturated_terminal_queues_preserve_every_byte_and_daemon_liveness` failed its queue saturation oracle. The fixture received `{"error":"terminal unavailable","reason":"resource_unavailable"}`; test result was 0 passed, 1 failed in 7.38s. Build-host load was 45.05/31.31/23.12. No infraternal/server sources are changed. I will retry after concurrent jobs clear.

kayg commented 2026-10-16 23:11:33 +00:00

Author

Owner

Copy link

The isolated agent_queues test passed on the same compiled binary: 1 passed, 0 failed, finished in 36.72s. The full gate remains red; I will retry it only after the other workspace runs clear so all standard gate rows pass together.

The isolated `agent_queues` test passed on the same compiled binary: 1 passed, 0 failed, finished in 36.72s. The full gate remains red; I will retry it only after the other workspace runs clear so all standard gate rows pass together.

kayg commented 2026-10-16 23:36:51 +00:00

Author

Owner

Copy link

Fourth iterate finding at ebd62f9: remote_runtime::tests::standby::control_queue::saturated_native_control_queue_does_not_hide_primary_terminal_output failed again (server group: 448 passed, 1 failed, 48 ignored; 293.78s). This run panicked at assertion failed: live_connection.close_reason().is_none() (crates/infraternal-server/src/remote_control_queue_tests.rs:216). The same exact test passed alone previously in 0.16s. Build-host load at gate completion was 16.58/19.08/20.54. I will rerun the full suite with RUST_TEST_THREADS=1 so every test runs without competing sibling tests.

Fourth iterate finding at `ebd62f9`: `remote_runtime::tests::standby::control_queue::saturated_native_control_queue_does_not_hide_primary_terminal_output` failed again (server group: 448 passed, 1 failed, 48 ignored; 293.78s). This run panicked at `assertion failed: live_connection.close_reason().is_none()` (`crates/infraternal-server/src/remote_control_queue_tests.rs:216`). The same exact test passed alone previously in 0.16s. Build-host load at gate completion was 16.58/19.08/20.54. I will rerun the full suite with `RUST_TEST_THREADS=1` so every test runs without competing sibling tests.

kayg commented 2026-10-16 23:37:04 +00:00

Author

Owner

Copy link

The control-queue test passed again in isolation on the same binary: 1 passed, 0 failed, 496 filtered out, finished in 0.10s. I will rerun the full iterate gate with RUST_TEST_THREADS=1 to keep every test while avoiding concurrent sibling tests in one binary.

The control-queue test passed again in isolation on the same binary: 1 passed, 0 failed, 496 filtered out, finished in 0.10s. I will rerun the full iterate gate with `RUST_TEST_THREADS=1` to keep every test while avoiding concurrent sibling tests in one binary.

kayg commented 2026-10-17 00:09:51 +00:00

Author

Owner

Copy link

At ebd62f98dd0493cf7a3f7b5f0cdcc5d705a15873, scripts/verify iterate passed with RUST_TEST_THREADS=1 to serialize the host-sensitive suite. The previously intermittent server control queue and terminal queue cases passed. Final output:

37 pass
0 fail
34463 expect() calls
Ran 37 tests across 10 files. [2.66s]

ITERATE gate passed. This is not the final gate: run scripts/verify fast before reporting done.

At `ebd62f98dd0493cf7a3f7b5f0cdcc5d705a15873`, `scripts/verify iterate` passed with `RUST_TEST_THREADS=1` to serialize the host-sensitive suite. The previously intermittent server control queue and terminal queue cases passed. Final output: 37 pass 0 fail 34463 expect() calls Ran 37 tests across 10 files. [2.66s] ITERATE gate passed. This is not the final gate: run scripts/verify fast before reporting done.

kayg commented 2026-10-17 00:15:34 +00:00

Author

Owner

Copy link

At ebd62f98dd0493cf7a3f7b5f0cdcc5d705a15873, the final trimmed vendor bundle passed the focused production browser suite on unique port 4287: 4 passed (1.4m). Wide final-glyph underline coverage was positive at DPR 1/2/3: lead/continuation accent pixels 7/11, 32/32, and 89/92. The theme test verified accent color, unchanged text color, hover/modifier show and hide, and both themes.

Fresh screenshots:

At `ebd62f98dd0493cf7a3f7b5f0cdcc5d705a15873`, the final trimmed vendor bundle passed the focused production browser suite on unique port 4287: 4 passed (1.4m). Wide final-glyph underline coverage was positive at DPR 1/2/3: lead/continuation accent pixels 7/11, 32/32, and 89/92. The theme test verified accent color, unchanged text color, hover/modifier show and hide, and both themes. Fresh screenshots:  

kayg commented 2026-10-17 00:28:34 +00:00

Author

Owner

Copy link

Vendor artifact check passed at ebd62f98dd0493cf7a3f7b5f0cdcc5d705a15873. I supplied the job scratch Zig 0.15.2 path via ZIG; the PATH-only attempt had exited before the build because Zig was not on PATH. The check reproduced the checked-in artifacts:

reproduced ghostty-web 5f0c6b7 with Ghostty 5714ed0 (Bun 1.4.2, Zig 0.15.2)

git diff --check passed.

Vendor artifact check passed at `ebd62f98dd0493cf7a3f7b5f0cdcc5d705a15873`. I supplied the job scratch Zig 0.15.2 path via `ZIG`; the PATH-only attempt had exited before the build because Zig was not on PATH. The check reproduced the checked-in artifacts: reproduced ghostty-web 5f0c6b7 with Ghostty 5714ed0 (Bun 1.4.2, Zig 0.15.2) git diff --check passed.

kayg commented 2026-10-17 00:28:51 +00:00

Author

Owner

Copy link

Committed the four-file critical-path trim as f25c5f0705c (Keep terminal link underlines within the critical transfer budget (#232)). It retains wide-glyph endpoint width while removing duplicate range clone/equality work; focused browser checks passed and build.sh --check reproduced the pinned artifacts.

Committed the four-file critical-path trim as `f25c5f0705c` (`Keep terminal link underlines within the critical transfer budget (#232)`). It retains wide-glyph endpoint width while removing duplicate range clone/equality work; focused browser checks passed and `build.sh --check` reproduced the pinned artifacts.

kayg commented 2026-10-17 00:29:07 +00:00

Author

Owner

Copy link

Pushed f25c5f0705c to origin/link-underline. The worktree was clean at push time.

Pushed `f25c5f0705c` to `origin/link-underline`. The worktree was clean at push time.

kayg commented 2026-10-17 00:41:37 +00:00

Author

Owner

Copy link

Post-commit production transfer measurement for merge base 41a3a81abceaa4ea7457d84ee05297ca2244efcc → f25c5f070b689496566f89f98a9663477bb7ed15:

• Critical JavaScript gzip: 297,678 → 297,952 B, +274 B (90 B under the +364 B cap).
• Critical JavaScript Brotli: 247,639 → 247,684 B, +45 B.
• Deferred JavaScript gzip: +29 B; on-demand JavaScript gzip: +393 B.
• All JavaScript classes gzip: 427,151 → 427,847 B, +696 B.

production-bundle-postcommit.log

Post-commit production transfer measurement for merge base `41a3a81abceaa4ea7457d84ee05297ca2244efcc` → `f25c5f070b689496566f89f98a9663477bb7ed15`: - Critical JavaScript gzip: 297,678 → 297,952 B, +274 B (90 B under the +364 B cap). - Critical JavaScript Brotli: 247,639 → 247,684 B, +45 B. - Deferred JavaScript gzip: +29 B; on-demand JavaScript gzip: +393 B. - All JavaScript classes gzip: 427,151 → 427,847 B, +696 B. [production-bundle-postcommit.log](https://git.kayg.org/attachments/ff92d1e5-418d-4fb0-9de3-6d8d915a468d)

kayg commented 2026-10-17 00:48:48 +00:00

Author

Owner

Copy link

Perf setup finding on infraternal-perf-test: with scrollback=40000000 in the URL, the origin/dev build ignored that test-only setting and reported configuredScrollback=10000, scrollbackRows=1341, rows=42 after the 10,000 LF-separated URL lines. That run produced no timing samples. I’m correcting the paired run to use the shared 10,000-row default on both builds and CRLF-delimited URL rows, then I’ll report the actual retained row counts from both buffer metrics and screen text.

perf-origin-dev-40m-request-default-retention.log

Perf setup finding on `infraternal-perf-test`: with `scrollback=40000000` in the URL, the `origin/dev` build ignored that test-only setting and reported `configuredScrollback=10000`, `scrollbackRows=1341`, `rows=42` after the 10,000 LF-separated URL lines. That run produced no timing samples. I’m correcting the paired run to use the shared 10,000-row default on both builds and CRLF-delimited URL rows, then I’ll report the actual retained row counts from both buffer metrics and screen text. [perf-origin-dev-40m-request-default-retention.log](https://git.kayg.org/attachments/dc845448-f02c-4edf-ab60-1ee22adc68df)

kayg commented 2026-10-17 00:55:25 +00:00

Author

Owner

Copy link

The corrected origin/dev run used the shared scrollback=10000 setting and CRLF-separated 10,000 URL lines (360,013 bytes). It reported configuredScrollback=10000, 1,119 retained scrollback rows / 1,160 URL rows, 42 viewport rows, 53 columns, and confirmed the final URL rendered. Load average at start was [1.57, 0.74, 0.46]. The 24 hover-to-link-cursor samples had p95 0.30 ms. The first scroll capture stalled after link_scan_ready, so no scroll-frame result was recorded. I stopped this run’s PIDs 2158082 and 2158103; the systemd scope is inactive and the lock is released.

perf-origin-dev-scroll-stall.log

The corrected `origin/dev` run used the shared `scrollback=10000` setting and CRLF-separated 10,000 URL lines (360,013 bytes). It reported `configuredScrollback=10000`, 1,119 retained scrollback rows / 1,160 URL rows, 42 viewport rows, 53 columns, and confirmed the final URL rendered. Load average at start was `[1.57, 0.74, 0.46]`. The 24 hover-to-link-cursor samples had p95 0.30 ms. The first scroll capture stalled after `link_scan_ready`, so no scroll-frame result was recorded. I stopped this run’s PIDs 2158082 and 2158103; the systemd scope is inactive and the lock is released. [perf-origin-dev-scroll-stall.log](https://git.kayg.org/attachments/781b6074-bd3a-4746-81cb-586f3e24c728)

kayg commented 2026-10-17 00:57:51 +00:00

Author

Owner

Copy link

I corrected the external perf harness so its frame sampler registers the requestAnimationFrame collector and returns before collection, reads viewport metrics from the product test API, and bounds each wheel command to 1 s. The paired baseline rerun uses fresh port 4392 and is queued behind the perf VM lock.

I corrected the external perf harness so its frame sampler registers the requestAnimationFrame collector and returns before collection, reads viewport metrics from the product test API, and bounds each wheel command to 1 s. The paired baseline rerun uses fresh port 4392 and is queued behind the perf VM lock.

kayg commented 2026-10-17 01:00:52 +00:00

Author

Owner

Copy link

The origin/dev run with 10,000-row default and CRLF URL rows reached the scroll phase, but alternating wheel direction on each event left the sampled viewport unchanged. No scroll p95 was produced. I changed the harness to keep one direction per 2 s run (starting upward from the bottom, then alternating direction between runs) and will retry on a fresh port.

perf-origin-dev-scroll-no-movement.log

The `origin/dev` run with 10,000-row default and CRLF URL rows reached the scroll phase, but alternating wheel direction on each event left the sampled viewport unchanged. No scroll p95 was produced. I changed the harness to keep one direction per 2 s run (starting upward from the bottom, then alternating direction between runs) and will retry on a fresh port. [perf-origin-dev-scroll-no-movement.log](https://git.kayg.org/attachments/33b44dbd-d660-4da8-ab50-aa4d4919dd8f)

kayg commented 2026-10-17 01:07:46 +00:00

Author

Owner

Copy link

Perf harness correction: the product canvas routes wheel input to canvas movement until the terminal input lock is enabled (D177; apps/tracer/src/canvas/camera/gestures.ts and apps/tracer/test/lock-icons.spec.ts). The no-movement sample therefore measured no terminal scrolling. The harness now activates the existing input-lock control through the UI, verifies aria-pressed=true, and uses fresh ports 4442/4443 for origin/dev and 4452/4453 for the branch. No product code change. Rerunning the paired 10,000-line scroll sample on the perf VM.

Perf harness correction: the product canvas routes wheel input to canvas movement until the terminal input lock is enabled (D177; apps/tracer/src/canvas/camera/gestures.ts and apps/tracer/test/lock-icons.spec.ts). The no-movement sample therefore measured no terminal scrolling. The harness now activates the existing input-lock control through the UI, verifies aria-pressed=true, and uses fresh ports 4442/4443 for origin/dev and 4452/4453 for the branch. No product code change. Rerunning the paired 10,000-line scroll sample on the perf VM.

kayg commented 2026-10-17 01:09:26 +00:00

Author

Owner

Copy link

Second perf harness correction: the input-lock toolbar button is present but hidden until the canvas toolbar row is rendered, so Playwright's visibility-checked click timed out. The existing lock-icons test activates the same button through DOM click after selecting the console window. I aligned the external harness with that established interaction; no product code change. Retrying on fresh ports 4462/4463 (origin/dev) and 4472/4473 (branch).

Second perf harness correction: the input-lock toolbar button is present but hidden until the canvas toolbar row is rendered, so Playwright's visibility-checked click timed out. The existing lock-icons test activates the same button through DOM click after selecting the console window. I aligned the external harness with that established interaction; no product code change. Retrying on fresh ports 4462/4463 (origin/dev) and 4472/4473 (branch).

kayg commented 2026-10-17 01:10:18 +00:00

Author

Owner

Copy link

Origin/dev baseline completed on infraternal-perf-test (4 vCPU, 8,262,776,688 bytes RAM), load average 0.72/1.74/1.33 at start and 0.91/1.76/1.33 at finish. Workload: 10,000 URL lines / 360,013 bytes; configured scrollback 10,000, retained scrollback 1,119 rows and screen URL rows 1,160. Hover-to-link-cursor p95 0.30 ms over 24 samples. Three scroll runs moved viewport; frame interval p95 16.8 ms in each run (run 1 p99 33.3 ms, max 50.1 ms; runs 2–3 p99/max 16.8 ms). Full raw log: https://git.kayg.org/attachments/b48ef74e-580b-4982-9086-4721789bfea1. Branch run follows against the same workload.

Origin/dev baseline completed on `infraternal-perf-test` (4 vCPU, 8,262,776,688 bytes RAM), load average 0.72/1.74/1.33 at start and 0.91/1.76/1.33 at finish. Workload: 10,000 URL lines / 360,013 bytes; configured scrollback 10,000, retained scrollback 1,119 rows and screen URL rows 1,160. Hover-to-link-cursor p95 0.30 ms over 24 samples. Three scroll runs moved viewport; frame interval p95 16.8 ms in each run (run 1 p99 33.3 ms, max 50.1 ms; runs 2–3 p99/max 16.8 ms). Full raw log: https://git.kayg.org/attachments/b48ef74e-580b-4982-9086-4721789bfea1. Branch run follows against the same workload.

kayg commented 2026-10-17 01:12:24 +00:00

Author

Owner

Copy link

Branch perf startup diagnostic: the candidate page rendered the plain TEST DOUBLE screen and never connected a product terminal. The served candidate main.js has the preview-only product test mode compiled out (useCanvas is false for mock=1), so this is a mismatched preview artifact, not an app startup regression. The candidate directory was reused for the production transfer-size comparison; I will rebuild a fresh preview artifact with INFRATERNALE_CLUSTERS_PREVIEW=1, re-stage it on the perf VM, and retry on fresh ports. No timing result from this run.

Branch perf startup diagnostic: the candidate page rendered the plain `TEST DOUBLE` screen and never connected a product terminal. The served candidate `main.js` has the preview-only product test mode compiled out (`useCanvas` is false for `mock=1`), so this is a mismatched preview artifact, not an app startup regression. The candidate directory was reused for the production transfer-size comparison; I will rebuild a fresh preview artifact with `INFRATERNALE_CLUSTERS_PREVIEW=1`, re-stage it on the perf VM, and retry on fresh ports. No timing result from this run.

kayg commented 2026-10-17 01:29:25 +00:00

Author

Owner

Copy link

Branch-head fast gate completed at f25c5f070b (exit 0; RUST_TEST_THREADS=1; full raw transcript: https://git.kayg.org/attachments/e271e054-1a2b-40b7-ab92-1657932975f2).

Gate summary, verbatim:

+---------------------------+--------------+
| Rust gate                 | Result       |
+---------------------------+--------------+
| fmt                       | PASS         |
| VT native fmt             | PASS         |
| CLI plugin compile-out    | PASS         |
| clippy                    | PASS         |
| test                      | PASS         |
| cargo-deny                | PASS         |
| static builds             | PASS         |
+---------------------------+--------------+

==> Vendored Ghostty artifact checksums
dist/ghostty-web.js: OK
dist/ghostty-vt.wasm: OK
dist/index.d.ts: OK

 37 pass
 0 fail
 34463 expect() calls
Ran 37 tests across 10 files. [3.59s]

Static-size rows: x86_64-unknown-linux-musl 21,422,096 B exact cap PASS; aarch64-unknown-linux-musl 19,594,856 B exact cap PASS. Proceeding with the fresh preview build and paired branch perf run.

Branch-head fast gate completed at f25c5f070b689496566f89f98a9663477bb7ed15 (exit 0; `RUST_TEST_THREADS=1`; full raw transcript: https://git.kayg.org/attachments/e271e054-1a2b-40b7-ab92-1657932975f2). Gate summary, verbatim: ``` +---------------------------+--------------+ | Rust gate | Result | +---------------------------+--------------+ | fmt | PASS | | VT native fmt | PASS | | CLI plugin compile-out | PASS | | clippy | PASS | | test | PASS | | cargo-deny | PASS | | static builds | PASS | +---------------------------+--------------+ ==> Vendored Ghostty artifact checksums dist/ghostty-web.js: OK dist/ghostty-vt.wasm: OK dist/index.d.ts: OK 37 pass 0 fail 34463 expect() calls Ran 37 tests across 10 files. [3.59s] ``` Static-size rows: x86_64-unknown-linux-musl 21,422,096 B exact cap PASS; aarch64-unknown-linux-musl 19,594,856 B exact cap PASS. Proceeding with the fresh preview build and paired branch perf run.

kayg commented 2026-10-17 01:34:25 +00:00

Author

Owner

Copy link

Correction to the prior preview-build note: the latest candidate artifact still had product test mode disabled because my build command used INFRATERNALE_CLUSTERS_PREVIEW (extra E); apps/tracer/build.ts reads INFRATERNAL_CLUSTERS_PREVIEW. The served artifact was therefore not a valid product preview, and no branch timings were collected. Rebuilding with the exact repository variable and switching to fresh candidate ports 4502/4503.

Correction to the prior preview-build note: the latest candidate artifact still had product test mode disabled because my build command used `INFRATERNALE_CLUSTERS_PREVIEW` (extra `E`); `apps/tracer/build.ts` reads `INFRATERNAL_CLUSTERS_PREVIEW`. The served artifact was therefore not a valid product preview, and no branch timings were collected. Rebuilding with the exact repository variable and switching to fresh candidate ports 4502/4503.

kayg commented 2026-10-17 01:35:36 +00:00

Author

Owner

Copy link

Branch perf completed at head f25c5f070b. Both variants ran on infraternal-perf-test (x86_64, 4 vCPU, 8,262,776,688 bytes RAM) with the same 10,000-URL-line, 360,013-byte workload and scrollback=10000 query setting. Both reported configuredScrollback=10000, retainedScrollbackRows=1119, retainedUrlRows=1160, and a 42-row × 53-column viewport. The 1,119 value is the actual retained scrollback count; the earlier 9,957 figure was incorrect.

Origin/dev (650472b987dbd1fe7192248dd1367d0766c5fb79): load average 0.72/1.74/1.33 at start, 0.91/1.76/1.33 at finish. Hover to link cursor (24 samples): p95 0.30 ms, p99 0.40 ms. Three moving scroll runs each had frame p95 16.8 ms; p99 was 33.3 ms on run 1 and 16.8 ms on runs 2–3. Raw log: https://git.kayg.org/attachments/b48ef74e-580b-4982-9086-4721789bfea1

Branch (f25c5f070b689496566f89f98a9663477bb7ed15): load average 0.34/0.09/0.28 at start, 0.88/0.21/0.32 at finish. Hover to underline pixel (24 samples): p95 1.20 ms, p99 1.90 ms. Three moving scroll runs each had frame p95 16.8 ms and p99 16.8 ms; run 1 had one 133.4 ms maximum interval, while runs 2–3 maxed at 16.8 ms. Raw log: https://git.kayg.org/attachments/58607c25-4a64-4865-aa5e-0fcaa10dba06

The hover endpoints differ by design: origin/dev has no underline pixel, so its measurement ends when link detection exposes the pointer cursor; the branch measurement ends when the accent underline pixels appear. Scroll p95 is unchanged. Critical JavaScript gzip growth remains +274 B against the +364 B limit. The screenshots and branch-head fast-gate transcript are attached above.

Branch perf completed at head f25c5f070b689496566f89f98a9663477bb7ed15. Both variants ran on `infraternal-perf-test` (x86_64, 4 vCPU, 8,262,776,688 bytes RAM) with the same 10,000-URL-line, 360,013-byte workload and `scrollback=10000` query setting. Both reported configuredScrollback=10000, retainedScrollbackRows=1119, retainedUrlRows=1160, and a 42-row × 53-column viewport. The 1,119 value is the actual retained scrollback count; the earlier 9,957 figure was incorrect. Origin/dev (`650472b987dbd1fe7192248dd1367d0766c5fb79`): load average 0.72/1.74/1.33 at start, 0.91/1.76/1.33 at finish. Hover to link cursor (24 samples): p95 0.30 ms, p99 0.40 ms. Three moving scroll runs each had frame p95 16.8 ms; p99 was 33.3 ms on run 1 and 16.8 ms on runs 2–3. Raw log: https://git.kayg.org/attachments/b48ef74e-580b-4982-9086-4721789bfea1 Branch (`f25c5f070b689496566f89f98a9663477bb7ed15`): load average 0.34/0.09/0.28 at start, 0.88/0.21/0.32 at finish. Hover to underline pixel (24 samples): p95 1.20 ms, p99 1.90 ms. Three moving scroll runs each had frame p95 16.8 ms and p99 16.8 ms; run 1 had one 133.4 ms maximum interval, while runs 2–3 maxed at 16.8 ms. Raw log: https://git.kayg.org/attachments/58607c25-4a64-4865-aa5e-0fcaa10dba06 The hover endpoints differ by design: origin/dev has no underline pixel, so its measurement ends when link detection exposes the pointer cursor; the branch measurement ends when the accent underline pixels appear. Scroll p95 is unchanged. Critical JavaScript gzip growth remains +274 B against the +364 B limit. The screenshots and branch-head fast-gate transcript are attached above.

kayg commented 2026-10-17 01:37:08 +00:00

Author

Owner

Copy link

Finished and pushed. HEAD: f25c5f070b (origin/link-underline matches). The issue now has the two review fixes, wide-glyph evidence, paired perf logs, both theme screenshots, bundle-size receipt, and full fast-gate transcript.

Performance result: on the perf VM, with 10,000 URL lines and scrollback=10000, both variants retained 1,119 scrollback rows (1,160 URL rows including screen rows). Origin/dev hover-to-link-cursor p95 was 0.30 ms; branch hover-to-underline-pixel p95 was 1.20 ms. All three scroll runs on both variants had 16.8 ms frame p95. Branch critical JavaScript gzip grew by 274 B, within the +364 B limit. Machine and load averages plus p99/max details are in the preceding issue comment and attached raw logs.

Fast gate output, verbatim:

+---------------------------+--------------+
| Rust gate                 | Result       |
+---------------------------+--------------+
| fmt                       | PASS         |
| VT native fmt             | PASS         |
| CLI plugin compile-out    | PASS         |
| clippy                    | PASS         |
| test                      | PASS         |
| cargo-deny                | PASS         |
| static builds             | PASS         |
+---------------------------+--------------+

==> Vendored Ghostty artifact checksums
dist/ghostty-web.js: OK
dist/ghostty-vt.wasm: OK
dist/index.d.ts: OK

 37 pass
 0 fail
 34463 expect() calls
Ran 37 tests across 10 files. [3.59s]

Full raw output: https://git.kayg.org/attachments/e271e054-1a2b-40b7-ab92-1657932975f2

Finished and pushed. HEAD: f25c5f070b689496566f89f98a9663477bb7ed15 (`origin/link-underline` matches). The issue now has the two review fixes, wide-glyph evidence, paired perf logs, both theme screenshots, bundle-size receipt, and full fast-gate transcript. Performance result: on the perf VM, with 10,000 URL lines and `scrollback=10000`, both variants retained 1,119 scrollback rows (1,160 URL rows including screen rows). Origin/dev hover-to-link-cursor p95 was 0.30 ms; branch hover-to-underline-pixel p95 was 1.20 ms. All three scroll runs on both variants had 16.8 ms frame p95. Branch critical JavaScript gzip grew by 274 B, within the +364 B limit. Machine and load averages plus p99/max details are in the preceding issue comment and attached raw logs. Fast gate output, verbatim: ``` +---------------------------+--------------+ | Rust gate | Result | +---------------------------+--------------+ | fmt | PASS | | VT native fmt | PASS | | CLI plugin compile-out | PASS | | clippy | PASS | | test | PASS | | cargo-deny | PASS | | static builds | PASS | +---------------------------+--------------+ ==> Vendored Ghostty artifact checksums dist/ghostty-web.js: OK dist/ghostty-vt.wasm: OK dist/index.d.ts: OK 37 pass 0 fail 34463 expect() calls Ran 37 tests across 10 files. [3.59s] ``` Full raw output: https://git.kayg.org/attachments/e271e054-1a2b-40b7-ab92-1657932975f2

kayg commented 2026-10-17 02:10:33 +00:00

Author

Owner

Copy link

Demo video of f25c5f070 (production build, test data): hover underline, Ctrl held underlines all visible links, scroll leaves no stale underline, light theme.

link-underline-demo.webm

Merge blocked only on a rebase onto dev (conflict with #228 tooltip placement in terminal-links.on-demand.ts); sent back to the job.

Separate gap noticed: URL detection ends at the first wide (CJK) character, e.g. https://cjk.example.test/路径/文档 links only up to /路. Not part of this change.

Demo video of f25c5f070 (production build, test data): hover underline, Ctrl held underlines all visible links, scroll leaves no stale underline, light theme. [link-underline-demo.webm](https://git.kayg.org/attachments/fcafb97b-a5fe-4e30-b98e-01f6b477a7b4) Merge blocked only on a rebase onto dev (conflict with #228 tooltip placement in terminal-links.on-demand.ts); sent back to the job. Separate gap noticed: URL detection ends at the first wide (CJK) character, e.g. `https://cjk.example.test/路径/文档` links only up to `/路`. Not part of this change.

kayg commented 2026-10-17 02:11:11 +00:00

Author

Owner

Copy link

Continuing #232 on link-underline, current HEAD f25c5f070b689496566f89f98a9663477bb7ed15, rebasing onto fetched origin/dev at c47a438be76e8cb1e5a5328bd190520d6039b33f. Worktree is clean. Read the full #232 and #214 comment histories; next I will verify the two committed review tests fail before applying them.

Continuing #232 on `link-underline`, current HEAD `f25c5f070b689496566f89f98a9663477bb7ed15`, rebasing onto fetched `origin/dev` at `c47a438be76e8cb1e5a5328bd190520d6039b33f`. Worktree is clean. Read the full #232 and #214 comment histories; next I will verify the two committed review tests fail before applying them.

kayg commented 2026-10-17 02:16:05 +00:00

Author

Owner

Copy link

Confirmed both reviewer regressions before the existing fixes at test snapshot c0183a899 (pre-fix base 7a902273d). The scroll unit suite failed 13/14: stale paint remained {startX:0,startY:0,endX:30,endY:0} after the viewport moved. The production wide-glyph Playwright check failed at DPR 1/2/3: lead pixels 7/32/89, continuation pixels 0/0/0 (minimums 4/8/12). Logs: scroll, wide.

Confirmed both reviewer regressions before the existing fixes at test snapshot `c0183a899` (pre-fix base `7a902273d`). The scroll unit suite failed 13/14: stale paint remained `{startX:0,startY:0,endX:30,endY:0}` after the viewport moved. The production wide-glyph Playwright check failed at DPR 1/2/3: lead pixels 7/32/89, continuation pixels 0/0/0 (minimums 4/8/12). Logs: [scroll](https://git.kayg.org/attachments/1342c49b-48ec-4ecc-af2a-be0914a2d935), [wide](https://git.kayg.org/attachments/03e9a5e8-d3d2-4fd8-98c3-388c9afa547b).

kayg commented 2026-10-17 02:28:40 +00:00

Author

Owner

Copy link

Rebase complete: link-underline is based on origin/dev c47a438be76e8cb1e5a5328bd190520d6039b33f, head 221657f53. Both link-tooltip placement/wrapping and accent underline behavior are retained; the scroll-clear path now hides through hideLinkTip(). During replay, bun run check passed, the focused scroll suite passed 14/14, and the merged tooltip/underline browser suite passed 5/5 at port 4232. Exact-head browser and gate reruns are in progress.

Rebase complete: `link-underline` is based on `origin/dev` `c47a438be76e8cb1e5a5328bd190520d6039b33f`, head `221657f53`. Both link-tooltip placement/wrapping and accent underline behavior are retained; the scroll-clear path now hides through `hideLinkTip()`. During replay, `bun run check` passed, the focused scroll suite passed 14/14, and the merged tooltip/underline browser suite passed 5/5 at port 4232. Exact-head browser and gate reruns are in progress.

kayg commented 2026-10-17 02:32:41 +00:00

Author

Owner

Copy link

Exact-head tracer checks passed at 221657f53: bun run check; focused link unit/property suites, 15 passed with 54,612 assertions; TRACER_TEST_PORT=4232 bun x playwright test test/terminal-links.spec.ts test/renderer-link-wide-review.spec.ts, 6 passed. The pixel test reports lead/continuation accent pixels 7/11, 32/32, 89/92 at DPR 1/2/3. Dev’s long-link tooltip checks passed at phone and desktop sizes; labels stayed within the tooltip and clear of the wrapped rows. Raw outputs: unit, browser.

Final ink/paper captures (393×852, 375×667, 1440×900):

Exact-head tracer checks passed at `221657f53`: `bun run check`; focused link unit/property suites, 15 passed with 54,612 assertions; `TRACER_TEST_PORT=4232 bun x playwright test test/terminal-links.spec.ts test/renderer-link-wide-review.spec.ts`, 6 passed. The pixel test reports lead/continuation accent pixels 7/11, 32/32, 89/92 at DPR 1/2/3. Dev’s long-link tooltip checks passed at phone and desktop sizes; labels stayed within the tooltip and clear of the wrapped rows. Raw outputs: [unit](https://git.kayg.org/attachments/4bddba0f-d921-4f00-8813-2ac5f17ccd65), [browser](https://git.kayg.org/attachments/08cb1006-9250-43b2-b33c-9555d526a096). Final ink/paper captures (393×852, 375×667, 1440×900):      

kayg commented 2026-10-17 03:01:35 +00:00

Author

Owner

Copy link

Production transfer-budget comparison completed at head 221657f535be94768f0e3285c00fcc2822cd80ee, against origin/dev c47a438be76e8cb1e5a5328bd190520d6039b33f.

Critical JavaScript: origin/dev 963,430 B identity / 298,149 B gzip / 247,962 B brotli; branch 963,816 B / 298,398 B / 248,239 B. Delta: +386 B identity, +249 B gzip, +277 B brotli. The critical-path gzip growth is within the +364 B limit. Both transfer-budget runs passed; largest on-demand chunk is 15,945 B gzip / 14,169 B brotli.

Receipts: origin/dev transfer report, branch transfer report.

Production transfer-budget comparison completed at head `221657f535be94768f0e3285c00fcc2822cd80ee`, against `origin/dev` `c47a438be76e8cb1e5a5328bd190520d6039b33f`. Critical JavaScript: origin/dev 963,430 B identity / 298,149 B gzip / 247,962 B brotli; branch 963,816 B / 298,398 B / 248,239 B. Delta: +386 B identity, +249 B gzip, +277 B brotli. The critical-path gzip growth is within the +364 B limit. Both transfer-budget runs passed; largest on-demand chunk is 15,945 B gzip / 14,169 B brotli. Receipts: [origin/dev transfer report](https://git.kayg.org/attachments/b372d673-58fd-493c-9002-279232c12a40), [branch transfer report](https://git.kayg.org/attachments/267a45f2-d58d-44f6-866f-fcd7a384fd09).

kayg commented 2026-10-17 03:13:42 +00:00

Author

Owner

Copy link

The shared origin/dev ref advanced to 06be8b0000f7d476eeeeed3472a4605757151243 during this job. git merge-base origin/dev link-underline remains c47a438be76e8cb1e5a5328bd190520d6039b33f, the base used for the rebase. I stopped the static baseline build as soon as it started against the moving ref; it produced no size measurement. The comparison is pinned to c47a438be76e8cb1e5a5328bd190520d6039b33f and will be rerun.

The shared `origin/dev` ref advanced to `06be8b0000f7d476eeeeed3472a4605757151243` during this job. `git merge-base origin/dev link-underline` remains `c47a438be76e8cb1e5a5328bd190520d6039b33f`, the base used for the rebase. I stopped the static baseline build as soon as it started against the moving ref; it produced no size measurement. The comparison is pinned to `c47a438be76e8cb1e5a5328bd190520d6039b33f` and will be rerun.

kayg commented 2026-10-17 03:15:57 +00:00

Author

Owner

Copy link

The first static size baseline attempt produced no binary: I invoked scripts/static-cargo without the repository's pinned compiler bootstrap. The build log shows .cargo/musl-cc.sh:53: .../target/auth-tools/zig/zig: No such file or directory while compiling openssl-sys. The worktree was restored to link-underline; I’m running scripts/auth-toolchain and repeating the baseline at the verified merge base c47a438be76e8cb1e5a5328bd190520d6039b33f.

Receipt: static-baseline.log

The first static size baseline attempt produced no binary: I invoked `scripts/static-cargo` without the repository's pinned compiler bootstrap. The build log shows `.cargo/musl-cc.sh:53: .../target/auth-tools/zig/zig: No such file or directory` while compiling `openssl-sys`. The worktree was restored to `link-underline`; I’m running `scripts/auth-toolchain` and repeating the baseline at the verified merge base `c47a438be76e8cb1e5a5328bd190520d6039b33f`. Receipt: [static-baseline.log](https://git.kayg.org/attachments/8e85c79b-3681-4ab0-824c-da28cba77dd7)

kayg commented 2026-10-17 03:19:20 +00:00

Author

Owner

Copy link

The first perf VM lock window ended without a measurement. The bundled mock server started from the job root, so its evidence-path guard classified /root/infraternal-bench/link-underline/evidence as inside its checkout and exited before serving. Server PID 2578712 exited with status 1 and the perf lock was released. I’m placing the bundle under a test-tree-shaped directory and setting the evidence path to a sibling of that checkout root before retrying.

Receipt: server startup error

The first perf VM lock window ended without a measurement. The bundled mock server started from the job root, so its evidence-path guard classified `/root/infraternal-bench/link-underline/evidence` as inside its checkout and exited before serving. Server PID 2578712 exited with status 1 and the perf lock was released. I’m placing the bundle under a test-tree-shaped directory and setting the evidence path to a sibling of that checkout root before retrying. Receipt: [server startup error](https://git.kayg.org/attachments/6b2c19ac-2ac1-4d43-b5c1-79b32b92f910)

kayg commented 2026-10-17 03:26:51 +00:00

Author

Owner

Copy link

The corrected origin/dev run reached the 10,000-line workload on infraternal-perf-test (x64, 4 CPUs, 8,262,776,688 bytes RAM) with scrollback=10000. It retained 1,483 scrollback rows / 1,520 URL rows including the visible screen (38 viewport rows × 44 columns; 360,013 output bytes). Under start load average 2.29/1.87/1.45, 24 pointer-to-link-cursor samples had p95 9.7 ms and p99 21.0 ms. No scroll frames were measured: Playwright timed out because the input-lock button is present but not visible to its click action. Server PID 2580121 exited with status 143 and the perf lock was released. I’m switching to the existing DOM click pattern and will repeat the paired baseline/candidate run.

Receipt: partial origin/dev run

The corrected origin/dev run reached the 10,000-line workload on `infraternal-perf-test` (x64, 4 CPUs, 8,262,776,688 bytes RAM) with `scrollback=10000`. It retained 1,483 scrollback rows / 1,520 URL rows including the visible screen (38 viewport rows × 44 columns; 360,013 output bytes). Under start load average 2.29/1.87/1.45, 24 pointer-to-link-cursor samples had p95 9.7 ms and p99 21.0 ms. No scroll frames were measured: Playwright timed out because the input-lock button is present but not visible to its click action. Server PID 2580121 exited with status 143 and the perf lock was released. I’m switching to the existing DOM click pattern and will repeat the paired baseline/candidate run. Receipt: [partial origin/dev run](https://git.kayg.org/attachments/42c2a32f-e102-4eef-812c-23bf8a54e63f)

kayg commented 2026-10-17 03:35:43 +00:00

Author

Owner

Copy link

Perf baseline run at c47a438be on the perf VM reached 10,000 URL lines (360,013 bytes), with configured scrollback 10,000, retained rows 1,483, viewport 38×44 and 37 visible links; load at start was 2.38/2.36/1.77. It stalled before hover samples. The harness waited on the hover promise before sending pointer movement; I fixed that sequencing and added bounded phase waits. No timing result is claimed from this run. Receipt: perf-origin-dev-final-rebased.log. The harness/server PIDs were stopped and the run's lock released.

Perf baseline run at c47a438be on the perf VM reached 10,000 URL lines (360,013 bytes), with configured scrollback 10,000, retained rows 1,483, viewport 38×44 and 37 visible links; load at start was 2.38/2.36/1.77. It stalled before hover samples. The harness waited on the hover promise before sending pointer movement; I fixed that sequencing and added bounded phase waits. No timing result is claimed from this run. Receipt: [perf-origin-dev-final-rebased.log](https://git.kayg.org/attachments/bc80d05f-2ccc-42bc-b814-a3dd722e9a31). The harness/server PIDs were stopped and the run's lock released.

kayg commented 2026-10-17 03:46:12 +00:00

Author

Owner

Copy link

Perf baseline completed on merge-base c47a438be (infraternal-perf-test, x64, 4 vCPU, 7.7 GiB). With scrollback=10000, the terminal retained 1,483 scrollback rows plus 38 viewport rows after 10,000 URL lines (360,013 bytes); 37 links were visible. Load average was 2.66/1.77/1.64 at start and 2.41/1.75/1.63 at end. Across 24 hovers, pointer-to-link-cursor p95 was 5.8 ms. Across three 2 s scroll runs, frame-interval p95 was 16.7 ms in each run. Receipt: origin-dev-perf-run.log. Candidate run follows with the same harness.

Perf baseline completed on merge-base c47a438be (`infraternal-perf-test`, x64, 4 vCPU, 7.7 GiB). With `scrollback=10000`, the terminal retained 1,483 scrollback rows plus 38 viewport rows after 10,000 URL lines (360,013 bytes); 37 links were visible. Load average was 2.66/1.77/1.64 at start and 2.41/1.75/1.63 at end. Across 24 hovers, pointer-to-link-cursor p95 was 5.8 ms. Across three 2 s scroll runs, frame-interval p95 was 16.7 ms in each run. Receipt: [origin-dev-perf-run.log](https://git.kayg.org/attachments/6d7b1f7d-36f0-4844-9624-24cd842f9549). Candidate run follows with the same harness.

kayg commented 2026-10-17 03:46:59 +00:00

Author

Owner

Copy link

D243 merge-base size measurement completed at c47a438be. Recipe: scripts/static-cargo build --locked --release --package infraternal --target x86_64-unknown-linux-musl, then the built ELF was confirmed static-pie and stripped. Size: 21,308,208 bytes. Receipt: static-baseline-merge-base.log. Candidate size will be taken from the final fast gate build.

D243 merge-base size measurement completed at c47a438be. Recipe: `scripts/static-cargo build --locked --release --package infraternal --target x86_64-unknown-linux-musl`, then the built ELF was confirmed static-pie and stripped. Size: 21,308,208 bytes. Receipt: [static-baseline-merge-base.log](https://git.kayg.org/attachments/7cf8218c-41b7-4c98-a227-b8c42f664bf3). Candidate size will be taken from the final fast gate build.

kayg commented 2026-10-17 03:47:33 +00:00

Author

Owner

Copy link

Candidate perf run reached the same 10,000-line workload (scrollback=10000, 1,483 retained rows, 38×44 viewport, 37 visible links) on infraternal-perf-test at load 2.87/2.01/1.73, but the first hover sample timed out before yielding a measurement. The server exited and the lock was released; no candidate timing result is claimed. I’m instrumenting the pointer-event bounds and canvas pixel check before retrying. Receipt: candidate-perf-run-failed.log.

Candidate perf run reached the same 10,000-line workload (`scrollback=10000`, 1,483 retained rows, 38×44 viewport, 37 visible links) on `infraternal-perf-test` at load 2.87/2.01/1.73, but the first hover sample timed out before yielding a measurement. The server exited and the lock was released; no candidate timing result is claimed. I’m instrumenting the pointer-event bounds and canvas pixel check before retrying. Receipt: [candidate-perf-run-failed.log](https://git.kayg.org/attachments/a29f546a-a085-4f19-92f9-19f6ed238e6d).

kayg commented 2026-10-17 03:49:32 +00:00

Author

Owner

Copy link

The candidate pointer diagnostic confirmed one matching event at (33, 122) on the canvas and cursor=pointer; the link was cached at column 0, visible row 0. The timeout came from the harness's exact-RGB pixel match, which missed the renderer's anti-aliased underline. I changed the measurement probe to use the accent-to-background pixel projection used by the browser assertions. No app change was made. Receipt: candidate-perf-diagnostic.log.

The candidate pointer diagnostic confirmed one matching event at (33, 122) on the canvas and `cursor=pointer`; the link was cached at column 0, visible row 0. The timeout came from the harness's exact-RGB pixel match, which missed the renderer's anti-aliased underline. I changed the measurement probe to use the accent-to-background pixel projection used by the browser assertions. No app change was made. Receipt: [candidate-perf-diagnostic.log](https://git.kayg.org/attachments/2b738201-50d0-4d20-b101-67dc0e410922).

kayg commented 2026-10-17 03:50:04 +00:00

Author

Owner

Copy link

Candidate perf completed on infraternal-perf-test (x64, 4 vCPU, 7.7 GiB), with the same 10,000 URL lines and scrollback=10000: 1,483 retained scrollback rows, 38 viewport rows and 37 visible links. At start/end load 3.45/2.17/1.80 and 3.24/2.16/1.81, pointer-to-accent-underline p95 was 9.7 ms over 24 hovers. Three 2 s scroll runs had frame-interval p95 16.8, 16.8 and 16.7 ms. The baseline run started at load 2.66/1.77/1.64, so the hover numbers are not a load-matched delta; scroll p95 stayed at 16.7–16.8 ms in both. Receipt: candidate-perf-run-final.log.

Candidate perf completed on `infraternal-perf-test` (x64, 4 vCPU, 7.7 GiB), with the same 10,000 URL lines and `scrollback=10000`: 1,483 retained scrollback rows, 38 viewport rows and 37 visible links. At start/end load 3.45/2.17/1.80 and 3.24/2.16/1.81, pointer-to-accent-underline p95 was 9.7 ms over 24 hovers. Three 2 s scroll runs had frame-interval p95 16.8, 16.8 and 16.7 ms. The baseline run started at load 2.66/1.77/1.64, so the hover numbers are not a load-matched delta; scroll p95 stayed at 16.7–16.8 ms in both. Receipt: [candidate-perf-run-final.log](https://git.kayg.org/attachments/c40ee85d-df58-4d3e-b8f7-26da7268908f).

kayg commented 2026-10-17 04:13:48 +00:00

Author

Owner

Copy link

The final gate's x86_64 static size check reports 21,308,208 bytes against the 21,300,016-byte cap (+8,192). The independently built c47a438be merge-base binary measured the same 21,308,208 bytes with SHA-256 7dfeac9024fc8706bbf2cbd9a5c35951c3ae2a3a10b5e90095c43c9a0e25217f; this lane changes no Rust files, so the measured binary delta is 0 bytes. The fast gate is continuing its remaining static checks; I’ll include the complete output at finish. Baseline receipt: static-baseline-merge-base.log.

The final gate's x86_64 static size check reports 21,308,208 bytes against the 21,300,016-byte cap (+8,192). The independently built c47a438be merge-base binary measured the same 21,308,208 bytes with SHA-256 `7dfeac9024fc8706bbf2cbd9a5c35951c3ae2a3a10b5e90095c43c9a0e25217f`; this lane changes no Rust files, so the measured binary delta is 0 bytes. The fast gate is continuing its remaining static checks; I’ll include the complete output at finish. Baseline receipt: [static-baseline-merge-base.log](https://git.kayg.org/attachments/7cf8218c-41b7-4c98-a227-b8c42f664bf3).

kayg commented 2026-10-17 04:25:56 +00:00

Author

Owner

Copy link

Finished and force-pushed link-underline with lease. Head: 221657f535be94768f0e3285c00fcc2822cd80ee (f25c5f070 → 221657f53). No deploy.

Rebase keeps the #228 tooltip placement/wrapping and #232 accent underline, modifier hold, synchronous scroll clearing, and wide-glyph continuation. Tracer unit checks: 15 pass, 0 fail, 54,612 expect() calls. TRACER_TEST_PORT=4232 bun x playwright test test/terminal-links.spec.ts test/renderer-link-wide-review.spec.ts: 6 passed; the tooltip suite: 5 passed. Screenshots: ink 393×852, paper 393×852, ink 375×667, paper 375×667, ink 1440×900, paper 1440×900.

Bundle budget passed: critical JS is 298,398 B gzip / 248,239 B Brotli; delta vs c47a438be is +249 B gzip / +277 B Brotli, under the +364 B gzip limit.

Perf VM (infraternal-perf-test, x64, 4 vCPU, 7.7 GiB), 10,000 URL lines, scrollback=10000: 1,483 retained scrollback rows plus 38 viewport rows, 37 visible links. Baseline load start 2.66/1.77/1.64: pointer-to-link-cursor p95 5.8 ms; scroll frame p95 16.7–16.7 ms. Candidate load start 3.45/2.17/1.80: pointer-to-accent-underline p95 9.7 ms; scroll frame p95 16.7–16.8 ms. Loads differ, so the hover readings are not a load-matched delta. Raw receipts are attached in the preceding perf comments.

D243 size: stripped x86_64 static binary is 21,308,208 bytes on both branch and merge base, delta 0 bytes; SHA-256 matches. The aarch64 binary is 19,508,568 bytes. No Rust files changed.

scripts/verify fast final output (full 325,692-byte transcript: verify-fast-final.log):

PT_INTERP: absent; size: 21308208 bytes; cap: 21300016 bytes; delta: +8192 bytes
binary exceeds the 21300016-byte size budget: 21308208 bytes
PT_INTERP: absent; size: 19508568 bytes; cap: 19500376 bytes; delta: +8192 bytes
binary exceeds the 19500376-byte size budget: 19508568 bytes
(fail) x86_64-unknown-linux-musl compiler metadata is independent of the checkout location [56.60ms]
(fail) aarch64-unknown-linux-musl compiler metadata is independent of the checkout location [38.04ms]
 0 pass
 2 fail
 10 expect() calls
Ran 2 tests across 1 file. [1.53s]
(static builds: 680s)

+---------------------------+--------------+
| Rust gate                 | Result       |
+---------------------------+--------------+
| fmt                       | PASS         |
| VT native fmt             | PASS         |
| CLI plugin compile-out    | PASS         |
| clippy                    | PASS         |
| test                      | PASS         |
| cargo-deny                | PASS         |
| static builds             | FAIL (1)     |
+---------------------------+--------------+

The static gate failures are recorded on this issue: both binaries exceed their current caps by 8,192 bytes, and the metadata checks found the Cargo registry path in both images. The x86_64 binary is byte-for-byte identical to the c47a438be merge-base artifact, so these failures are present in the base as well as this branch.

Finished and force-pushed `link-underline` with lease. Head: `221657f535be94768f0e3285c00fcc2822cd80ee` (`f25c5f070` → `221657f53`). No deploy. Rebase keeps the #228 tooltip placement/wrapping and #232 accent underline, modifier hold, synchronous scroll clearing, and wide-glyph continuation. Tracer unit checks: `15 pass`, `0 fail`, `54,612 expect() calls`. `TRACER_TEST_PORT=4232 bun x playwright test test/terminal-links.spec.ts test/renderer-link-wide-review.spec.ts`: `6 passed`; the tooltip suite: `5 passed`. Screenshots: [ink 393×852](https://git.kayg.org/attachments/6380c086-5388-4e6b-a442-523feabe31fa), [paper 393×852](https://git.kayg.org/attachments/882f853b-3126-4c8b-9f20-cbc5f784fdce), [ink 375×667](https://git.kayg.org/attachments/29bb63bd-fe6a-4fd0-ad19-ccc12f28cf01), [paper 375×667](https://git.kayg.org/attachments/6da2732d-3a26-4358-bc09-b57fa670fb4a), [ink 1440×900](https://git.kayg.org/attachments/c409eea2-601d-4840-a1c9-81292729a3c2), [paper 1440×900](https://git.kayg.org/attachments/c703b7be-da3e-44aa-9ca9-fa0b0f4cbdf0). Bundle budget passed: critical JS is 298,398 B gzip / 248,239 B Brotli; delta vs c47a438be is +249 B gzip / +277 B Brotli, under the +364 B gzip limit. Perf VM (`infraternal-perf-test`, x64, 4 vCPU, 7.7 GiB), 10,000 URL lines, `scrollback=10000`: 1,483 retained scrollback rows plus 38 viewport rows, 37 visible links. Baseline load start 2.66/1.77/1.64: pointer-to-link-cursor p95 5.8 ms; scroll frame p95 16.7–16.7 ms. Candidate load start 3.45/2.17/1.80: pointer-to-accent-underline p95 9.7 ms; scroll frame p95 16.7–16.8 ms. Loads differ, so the hover readings are not a load-matched delta. Raw receipts are attached in the preceding perf comments. D243 size: stripped x86_64 static binary is 21,308,208 bytes on both branch and merge base, delta 0 bytes; SHA-256 matches. The aarch64 binary is 19,508,568 bytes. No Rust files changed. `scripts/verify fast` final output (full 325,692-byte transcript: [verify-fast-final.log](https://git.kayg.org/attachments/87882b9a-8d2d-46c9-b17c-6d612fdf6d5f)): ```text PT_INTERP: absent; size: 21308208 bytes; cap: 21300016 bytes; delta: +8192 bytes binary exceeds the 21300016-byte size budget: 21308208 bytes PT_INTERP: absent; size: 19508568 bytes; cap: 19500376 bytes; delta: +8192 bytes binary exceeds the 19500376-byte size budget: 19508568 bytes (fail) x86_64-unknown-linux-musl compiler metadata is independent of the checkout location [56.60ms] (fail) aarch64-unknown-linux-musl compiler metadata is independent of the checkout location [38.04ms] 0 pass 2 fail 10 expect() calls Ran 2 tests across 1 file. [1.53s] (static builds: 680s) +---------------------------+--------------+ | Rust gate | Result | +---------------------------+--------------+ | fmt | PASS | | VT native fmt | PASS | | CLI plugin compile-out | PASS | | clippy | PASS | | test | PASS | | cargo-deny | PASS | | static builds | FAIL (1) | +---------------------------+--------------+ ``` The static gate failures are recorded on this issue: both binaries exceed their current caps by 8,192 bytes, and the metadata checks found the Cargo registry path in both images. The x86_64 binary is byte-for-byte identical to the c47a438be merge-base artifact, so these failures are present in the base as well as this branch.

kayg commented 2026-10-17 04:29:16 +00:00

Author

Owner

Copy link

Merged to dev as f627a2e09 and deployed web-only to production (no restart). release-prod-live = f627a2e09. Owner saw the demo video. Follow-ups queued: URL detection stops at the first wide (CJK) character; last text column possibly clipped while the scrollbar shows.

Merged to dev as f627a2e09 and deployed web-only to production (no restart). release-prod-live = f627a2e09. Owner saw the demo video. Follow-ups queued: URL detection stops at the first wide (CJK) character; last text column possibly clipped while the scrollbar shows.

kayg referenced this issue 2026-10-28 17:53:18 +00:00

Terminal: recognise links and make them clickable (incl. wrapped URLs and OSC 8) #214

kayg referenced this issue 2026-10-28 17:53:55 +00:00

Orchestrator handover 2026-10-28: moving to a new machine #254

kayg referenced this issue 2026-10-28 17:53:56 +00:00

Orchestrator handover 2026-10-28: moving to a new machine #254

kayg referenced this issue 2026-10-28 18:00:45 +00:00

Orchestrator handover 2026-10-28: moving to a new machine #254

kayg commented 2026-11-08 19:04:28 +00:00

Author

Owner

Copy link

Closed by the orchestrator during the 2026-11-09 night housekeeping (owner order: "close stale prs/issues"). The work has landed on dev, and the latest note here is a finish or merge note (J387 inventory on #436). Reopen if anything is still missing.

Closed by the orchestrator during the 2026-11-09 night housekeeping (owner order: "close stale prs/issues"). The work has landed on dev, and the latest note here is a finish or merge note (J387 inventory on #436). Reopen if anything is still missing.

kayg closed this issue 2026-11-08 19:04:28 +00:00

Sign in to join this conversation.

No Branch/Tag specified

Branches Tags

main

j470b-server-mesh-rejoin

mesh-compat-382

j444b-d381-key-wins-flood

j461b-macos-utun-reader-thread

j475b-remove-host-motion

fix-reconnect-new-link

j481-d380

dev

j479-mesh-soak

fix/j476-host-facts

j475-motion-pill-split

j470-mac-node-join

root-upgrade-d376

j474-review-merge

latency-targets-120hz

nightly-suite-shards

gotatun-pin-j469

macos-utun-reader-thread

fix-flood-echo-stall

wip/j416

motion-fast-ease-out

mesh-addhost-j259

takeover-close-terminals-d368

kbd-instant-followups

fix-trust-key-double-click

nightly-suite-partial-report

appwin-icons-j207

rusqlite-slice1

swipe-content-first-d374

mesh-mtu-381

addhost-instant-open

glass-contrast-d365-mock

glass-contrast-d365

signout-followups-442

fix-startup-toast-red

research-gotatun-matrix-j317

themes-sync-j304

fix-load-flakes

fix-nix-zig-patch

fix-compute01-browser-baseline

ratchet-size-baseline-419

predict-auto-d130

revoke-coverage-d367

nightly-suite-queue-timeout

nightly-browser-suite

fix-rustsec-2026-0459

fonts-ioskeley-obel

log-retention-d375

fix-kbd-no-anim-427

fix-dev-red-specs-2

takeover-proof-305-j205

wip/j410-server-pump

ui-consistency-d372

fix-osc52-clipboard

libghostty-vt-1.0

forget-retired-host-d369

j404-merge-426

fix-setup-lock-race

fix-touch-select-426

revoke-access-report-ui

fix-predictive-echo-specs

fix-selinux-agent-exec

wip/j395-checkpoint2

wip/j395

j391-review-scratch

ui-consistency-j195-deferred-rebased

fix-rustc-paths-421

decisions-wording-341

fix-install-suite-red

fix-dev-rtt-specs

fix-pending-removal-align

fix-privileged-skip

fix-install-suite-404

review-j368

mesh-dns-j254

revoke-access-report-d356b

revoke-privilege-round5

fix-removal-confirm

review-j360

gotatun-rotation-367

deps-refresh-340

review-j354

review-J351

review-j350

addhost-takeover-d359

revoke-report-items-347

revoke-end-sessions-d362

review-j348

size-mesh-ssh-346

size-trim-337

revoke-access-diff-d356

j259-pre-rebase-b06c5a6

j305-ended-at-start-spec

addhost-sudo-consent

fix-layout-real-specs-411

fix-shell-open-411

fix-console-hosts-373

fix-shell-exit

fuzz-host-name-353

fix-host-details-410

macos-filter-j197

research-gotatun-tricks

revoke-p3-j216l

stale-state-pill

revoke-followups-j216k

revoke-progress-j216e

backup/macos01-20261106/lume/j197

backup/macos01-20261106/lume/appwin-icons-j207-stash0

backup/macos01-20261106/claude/stash-0

backup/macos01-20261106/claude/stash-1

backup/macos01-20261106/claude/J80-macos-s1

backup/macos01-20261106/claude/J80

backup/macos01-20261106/claude/J80-slice1-20261102

backup/macos01-20261106/claude/J16b-appwin-on-dev

backup/macos01-20261106/claude/appwin-launch-idle-146

backup/macos01-20261106/claude/J80-s1-run

wip/j259

trust-285-p3c

runtime-lock-j216f

mesh-ssh-p3-j175

tun-proof-p3b

trust-285-p3

terminal-polish-j271

tun-proof-p3

ssh-install-flakes-j216i

fuzz-fixes-j264

trust-285-a

trust-285-b

restart-oversized-fixture-j198

backup/macos01-20261106/lume/appwin-icons-j207

restart-bound-j216h

up-preflight-p3-j198

backup/macos01-20261106/lume/j251

up-preflight-j198

wip/up-preflight-j198-j245-p2

wip/up-preflight-j198-j245-p1

agent-fsync-offloop-j196

wip/up-preflight-j198-j245

settings-offscreen-j231

wip/up-preflight-j198-j226-inspection

wip/j103f

themes-j206

wip/up-preflight-j198-j226

flakes-j216c

backup/macos01-20261106/claude/j197

flakes-j216b

backup/macos01-20261106/claude/j197-red

wip/up-preflight-j198-review

flakes-j216

backup/macos01-20261106/claude/appwin-icons-j207

engineering-j208b

appwin-mock-j209

toast-stack-j165

engineering-j208

decisions-d351

decisions-d350

decisions-d349

appearance-presets-mock-j129

decisions-d348

docs-audit-spec

doc-coverage-baseline-raise

decisions-d347

decisions-d346b

decisions-d346

j88h-profile

ui-consistency-j195-deferred

ui-consistency-j195

engineering-practices-j194

preferred-login-ui-j166

signin-pools-263-j174

macos-node-265-s1

decisions-d345

decisions-d344

proxy-cgroup-j164

verify-quick-fuzz-329

proxy-start-io-j168

timing-flakes-j163

up-263-s1

mcp-events-j147

verify-quick-fuzz-j158

proxy-p3-j159

agent-resource-j157-detect

wip/j103d-final

remote-cancel-hang-j155

proxy-traefik-281-j62d

j149-verify-quick

proxy-traefik-281-j62b

j150-drop-unnamed

wip/j79d-red-r4

wip/j103d-d332

rootless-toast-j142

wip/j103d-head

wip/j103d

j125c-mcp

macos-build-j136

backup/macos01-20261106/claude/j136

wip/j79d-red-cli

wip/j79d-red-server

j121-link-liveness

j118c-tracer-failures

wip/j79c-head

release-desktop-j131

j124c-signed-macos

wip/j79c-red-p3

wip/j79c-red-p23

wip/j79c-red-p22

merge-j128-check

perf-deferred-j128

sidebar-toggle-260

j127-shortcut-changed-dot

j125-mcp

j118-tracer-failures

j126-account-summary

settings-sections-mock-j92

j120-pinch-lock

sidebar-292

backup/macos01-20261106/claude/j120

taste-2026-11-03

row-actions-242-215

backup/macos01-20261106/claude/stash-0-untracked

backup/macos01-20261106/claude/J80-265-s1-run

backup/macos01-20261106/claude/J80-macos-s1-20261102

j112-canvas-commands

ui-no-overshoot-scroll

fix-recovery-restart

backup/macos01-20261106/claude/stash-1-untracked

taste-2026-11-02

size-baseline-ratchet

machine-cards-structure-j108

j109-boot-copy-mockups

fix-dev-red

wip/up-263-s1

j95b-settings-rows

proxy-traefik-281

j95-option-rows

docs-human-j91

docs-keys-palette

fix-flood-1013-270-j38d

mesh-ssh-s1

mesh-ssh-s0

addhost-d291-tunnel

test-stability-load

mesh-relay-220

repro-296

shim-thp-off-24

mesh-pins-258-v2

mesh-dns

signin-d311-merged

docs-pages

ui-size-283

fix-ui-batch-specs

wip/j12b-edge-update

gate-on-arm

fix-rename-sync-293

trust-285-j60b

fix-focus-ring-clip

integ-ui-1101

edge-update-v2

signin-d311

taste-2026-11-01

layout-per-tab-250

ui-approved-splitopt-224

ui-approved-splitopt-224-on-dev

icon-text-centre-282

ui-approved-189-212-216

research-reverse-proxy

transfer-budget-320

review-mesh-pins-258-r2

sidebar-pill-259

review-mesh-relay-220

wip/j12-edge-update-v2

nix-package

status-crash-loop-237

fix-dev-test-failures

fix-fixture-socket-path

wip/mesh-relay-220-pre-dev-advance-rebase

fix-canvas-side-clip

research-head-to-head

frosted-real-256

review-mesh-pins-258

wip/mesh-relay-220-pre-final-rebase

review-layout-per-tab-250-r2

broad-suite-triage

nix-devshell

review-status-crash-loop-237-r3

wip/nix-devshell-pre-p3

taste-frosted-real

wip/code-serve-demo-2026-10-30

wip/code-serve-glass-2026-10-30

wip/code-serve-clusters-2026-10-30

wip/code-merge-release94-dev-2026-10-30

wip/rproxy-delta-scratch-2026-10-30

wip/stale-note-257-staged-2026-10-30

wip/d241-size-staged-2026-10-30

review-broad-suite-207

wip/mesh-pins-258-pre-split

wip/mesh-relay-220-pre-rebase

mesh-measure-220

plan-211-update

review-edge-autoupdate-r3

perf24-reconnect

review-stale-note-r2

release-prod-live

d241-shell-marker

review-mesh-pins-server-258

review-revocation-r3

review-setup-token-r2

review-sync-journal-r3

wip/review-mesh-pins-258-prune-review-1995fee46

frosted-scroll-256

sync-journal-r3

revocation-per-account-wip

review-tokio-asyncfd-0386

review-sync-p3s

wip/stash-8by64-stash0-2026-10-30

mesh-drag-r2

setup-token-data-wip

release-9-4

auth-busy-1013

mesh-view-overlap

sync-p3s

review-auth-busy-wake

review-readd-r8-wip

mesh-drag-bounds

audit-auth-93

app-windows-stream-start

review-authority-cap-r3

app-windows-r7-measure

readd-minimal-r6

readd-minimal-r5

app-windows-r7-soak

readd-minimal-r4

app-windows-r7

app-windows-r6

readd-minimal-r3

readd-minimal-r2

readd-minimal

merge-release8-dev

review-busy-page-liveness-r2

review-busy-page-liveness

mesh-view-trim

review-authority-read-retry-r2

review-remove-uninstall-r2

release-retry-reattach

release-retry

review-remove-uninstall

review-authority-read-retry

app-windows-ship

mesh-autoswitch-p2

mesh-ssh

mesh-relay

mesh-macos-node

screen-and-control

session-labels

review-real-logins-r5

review-real-logins-r4

review-real-logins-r3

shim-memory-r2

review-real-logins-r2

shim-memory

shadow-addhost-instant

shadow-real-logins

shadow-same-session-flood

shadow-prod-perf-fix

hosts-from-server

app-windows-real

app-windows-transport

release-prod-6w

mesh-policy-verify

quiet-close-toast

dev-merge-r4

review-mesh-agent-transport

review-app-windows-transport

settings-roundtrip

review-release-pipeline-r4

mesh-agent-transport

ended-strip-fix

review-audit-ca-update-d244

settings-rows

phone-width

signout-clear

fix-restart-tests

dev-merge-rp

signer-hardening

review-audit-ca-update

terminal-fidelity

review-audit-ca-update-wip

mesh-integrate

review-mesh-agent-transport-515d3712

review-mesh-dataplane-feec1988

mesh-dataplane

signout

research-canvas-node

wip/review-renderer-cost

wip/review-terminal-fidelity

padlock-glyph

review-release-pipeline-2e95b689

phone-polish

sidebar-polish

inbox-generic

review-mesh-dataplane-378d4055

prod-batch

ui-inbox

ui-ca-unlock

ui-machine-sections

review-mesh-dataplane-a5f744fb

tls-session-prod

session-bounds-prod

fix-dead-session

session-bounds

tls-session

palette-polish

shortcuts-d222

research-appmeta

phone-gutter

ui-centre-button

wip/ui-centre-button

wip/tls-session

review-addhost-breakit

ui-platform

renderer-cost

remote-agents

mesh-api

ssh-bootstrap

prod-hotfix2

prod-fixes

review-remote-agents

review-mesh-dataplane

session-life

research-browser

review-mesh-api

research-capture

review-mesh-api-round3

review-mesh-api-round1

gate-stability

review-gate-stability

review-relay-direct

review-mesh-api-findings

proto-sessions

font-picker

proto-addhost

proto-windows

mesh-core

latency-gate

research-bsd

research-discovery

keytray-mods

mock-core

ui-tweaks

fix-motion-sel

fix-motion

integrate2

fix-robust

t-web

t-server

vt-ghostty

fix-snap-echo

t-web-auth

review-vt2

t-agent

lane-protocol

lane-shim

lane-auth

s1-base

s1-contracts

s1-workspace

No results found.

Labels

Clear labels   

launch-qa

Findings from launch QA.

No labels

launch-qa

Milestone

Clear milestone

No items

No milestone

Projects

Clear projects

No items

No project

Assignees

Clear assignees

No assignees

1 participant

Notifications

Subscribe

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference

kayg/infraternal#232

No description provided.