kayg/infraternal
Watch
1
0
Fork
You've already forked infraternal
 0
Code Issues 356 Pull requests Projects Releases Packages Wiki Activity Actions

UI consistency audit (owner, 2026-11-04) #345

New issue
Open
opened 2026-11-04 10:03:59 +00:00 by kayg · 20 comments
kayg commented 2026-11-04 10:03:59 +00:00
Owner
Copy link

Owner request (2026-11-04): "do that consistency audit to see where things like text sizes, ui size, corner radius, etc differ and fix them if consistent and document if we decided it to be different."

Job J195. Branch ui-consistency-j195, base origin/dev a89ebb0ae.

Plan (estimate ~4 h):

  1. Audit: inventory every font, radius, spacing, control height, border, shadow, icon, z-index, motion and UI-size-scaling declaration in apps/tracer and packages/ui; compare to tokens.
  2. Classify each deviation: (a) accidental, fix; (b) decided, comment citing the decision; (c) unclear, owner question.
  3. Fix (a) in atomic commits per property. Skip files touched by open branches (deferred list posted here).
  4. Before/after screenshots from a production build (1440x900, 393x852, 375x667; dark and light) as a contact sheet here.
  5. Gate: verify quick, then scripts/verify fast, verbatim.
Owner request (2026-11-04): "do that consistency audit to see where things like text sizes, ui size, corner radius, etc differ and fix them if consistent and document if we decided it to be different." Job J195. Branch `ui-consistency-j195`, base origin/dev a89ebb0ae. Plan (estimate ~4 h): 1. Audit: inventory every font, radius, spacing, control height, border, shadow, icon, z-index, motion and UI-size-scaling declaration in `apps/tracer` and `packages/ui`; compare to tokens. 2. Classify each deviation: (a) accidental, fix; (b) decided, comment citing the decision; (c) unclear, owner question. 3. Fix (a) in atomic commits per property. Skip files touched by open branches (deferred list posted here). 4. Before/after screenshots from a production build (1440x900, 393x852, 375x667; dark and light) as a contact sheet here. 5. Gate: `verify quick`, then `scripts/verify fast`, verbatim.
kayg commented 2026-11-04 10:14:34 +00:00
Author
Owner
Copy link

Progress (J195) — branch ui-consistency-j195 (base origin/dev a89ebb0ae), stacked ui-consistency-j195-deferred.

Pushed so far:

  • ui-consistency-audit.spec.ts: captures + computed-style census of 6 screens x 3 sizes x ink/paper (off unless UI_AUDIT=1).
  • scripts/verify.d/ui-literals.py + a verify quick row: a change may not ADD px font sizes, px radii (>2 px) or pasted cubic-bezier(0.23,1,0.32,1). Diffs against the merge base (no baseline file), 0.07 s, opt-out by a ui-literal: <decision> comment.
  • doc.css: shared easing token + rem lengths (free files).

Deferred set is large. Branches active in the last 48 h (not wip/backup/mockup) touch 423 files under apps/tracer + packages/ui, i.e. every stylesheet except doc.css/docs*.css/data-toggle.css and ~90 % of all style declarations (3390 scanned; 3023 in deferred files). Fixes for those are in the stacked branch ui-consistency-j195-deferred (5 pixel-identical commits), NOT for merging until j129 / j132 / j193 / j166 land.

Finding (to be confirmed by the census): --ease is defined only on .auth-screen and the standalone docs pages, but packages/ui/styles.css (.check, .swap copy-tick morph D103, .glitch) reads var(--ease) with no fallback. Inside the app .terminal-screen defines --ease-out, not --ease, so those transition declarations would be invalid at computed time and the animation silently dropped. Fix in the deferred branch.

Queue: build-slot on compute01 is at position ~16, my run (before / after / after-deferred in one slot) waits there.

**Progress (J195)** — branch `ui-consistency-j195` (base origin/dev a89ebb0ae), stacked `ui-consistency-j195-deferred`. Pushed so far: - `ui-consistency-audit.spec.ts`: captures + computed-style census of 6 screens x 3 sizes x ink/paper (off unless `UI_AUDIT=1`). - `scripts/verify.d/ui-literals.py` + a `verify quick` row: a change may not ADD px font sizes, px radii (>2 px) or pasted `cubic-bezier(0.23,1,0.32,1)`. Diffs against the merge base (no baseline file), 0.07 s, opt-out by a `ui-literal: <decision>` comment. - doc.css: shared easing token + rem lengths (free files). **Deferred set is large.** Branches active in the last 48 h (not wip/backup/mockup) touch 423 files under `apps/tracer` + `packages/ui`, i.e. every stylesheet except `doc.css`/`docs*.css`/`data-toggle.css` and ~90 % of all style declarations (3390 scanned; 3023 in deferred files). Fixes for those are in the stacked branch `ui-consistency-j195-deferred` (5 pixel-identical commits), NOT for merging until j129 / j132 / j193 / j166 land. **Finding (to be confirmed by the census):** `--ease` is defined only on `.auth-screen` and the standalone docs pages, but `packages/ui/styles.css` (`.check`, `.swap` copy-tick morph D103, `.glitch`) reads `var(--ease)` with no fallback. Inside the app `.terminal-screen` defines `--ease-out`, not `--ease`, so those `transition` declarations would be invalid at computed time and the animation silently dropped. Fix in the deferred branch. **Queue:** build-slot on compute01 is at position ~16, my run (before / after / after-deferred in one slot) waits there.
kayg commented 2026-11-04 12:11:47 +00:00
Author
Owner
Copy link

Finding 1 (confirmed in a production build): shared components lose their transition inside the app.

packages/ui/styles.css .check (checkbox press), .swap (the copy-tick morph, D103) and .glitch read var(--ease). --ease is defined only on .auth-screen and the standalone docs pages; .terminal-screen defines --ease-out. A transition that reads an undefined variable is invalid at computed-value time, so the browser uses the initial value.

Probe: ui-consistency-audit.spec.ts builds <label class="check"> and <span class="swap"><svg/></span> inside each screen's root and reads getComputedStyle().transition* (origin/dev a89ebb0ae + the spec, 1440x900 ink):

screen .check .swap svg
canvas `0s ease`
sidebar `0s ease`
palette `0s ease`
window popover `0s ease`
sign-in `0.14s cubic-bezier(0.23, 1, 0.32, 1)`

So in the app the checkbox press and the copy-icon morph snap; on the sign-in card the same components animate. Fix is on ui-consistency-j195-deferred (fix: shared components read --ease-out), deferred because packages/ui/styles.css and auth.css are touched by open branches. Re-run of the probe on that branch follows.

Finding 2 (rendered values, 1440x900 ink, five screens, before any change):

  • font sizes: 15.4, 14.85, 14.3, 13.2, 12.1, 11 px (+ 23.1 and 16.5 on sign-in): eight sizes.
  • font weights: 450, 500, 550 (123 elements), 600, 650 (19): five, and TASTE §2 lists 450 / 500 / 600 / 400. 550 and 650 are the legacy --fw-label and --fw-title.
  • corner radii of the floating surfaces: sign-in card 24.2 px, window and sidebar 19.8 px, palette 17.6 px, window popover 15.4 px: four radii for "a floating card".

These are owner questions 1 and 2 in docs/design/UI-CONSISTENCY-AUDIT.md; nothing was changed for them.

**Finding 1 (confirmed in a production build): shared components lose their transition inside the app.** `packages/ui/styles.css` `.check` (checkbox press), `.swap` (the copy-tick morph, D103) and `.glitch` read `var(--ease)`. `--ease` is defined only on `.auth-screen` and the standalone docs pages; `.terminal-screen` defines `--ease-out`. A `transition` that reads an undefined variable is invalid at computed-value time, so the browser uses the initial value. Probe: `ui-consistency-audit.spec.ts` builds `<label class="check">` and `<span class="swap"><svg/></span>` inside each screen's root and reads `getComputedStyle().transition*` (origin/dev a89ebb0ae + the spec, 1440x900 ink): | screen | `.check` | `.swap svg` | |---|---|---| | canvas | `0s | ease` | `0s | ease` | | sidebar | `0s | ease` | `0s | ease` | | palette | `0s | ease` | `0s | ease` | | window popover | `0s | ease` | `0s | ease` | | sign-in | `0.14s | cubic-bezier(0.23, 1, 0.32, 1)` | `0.15s ... cubic-bezier(0.23, 1, 0.32, 1)` | So in the app the checkbox press and the copy-icon morph snap; on the sign-in card the same components animate. Fix is on `ui-consistency-j195-deferred` (`fix: shared components read --ease-out`), deferred because `packages/ui/styles.css` and `auth.css` are touched by open branches. Re-run of the probe on that branch follows. **Finding 2 (rendered values, 1440x900 ink, five screens, before any change):** - font sizes: 15.4, 14.85, 14.3, 13.2, 12.1, 11 px (+ 23.1 and 16.5 on sign-in): eight sizes. - font weights: 450, 500, 550 (123 elements), 600, 650 (19): five, and TASTE §2 lists 450 / 500 / 600 / 400. 550 and 650 are the legacy `--fw-label` and `--fw-title`. - corner radii of the floating surfaces: sign-in card 24.2 px, window and sidebar 19.8 px, palette 17.6 px, window popover 15.4 px: four radii for "a floating card". These are owner questions 1 and 2 in `docs/design/UI-CONSISTENCY-AUDIT.md`; nothing was changed for them.
kayg commented 2026-11-04 12:41:18 +00:00
Author
Owner
Copy link

Contact sheets and regression proof (J195) — production build, 36 captures per phase (6 screens x 3 sizes x ink/paper). Phases: before = origin/dev a89ebb0ae + spec (941569815), main = ui-consistency-j195, deferred = ui-consistency-j195-deferred (7 more commits, not for merge yet). Each sheet: rows canvas, sidebar, palette, settings, popover, sign-in; columns ink before, ink after, paper before, paper after. Attached below: before vs deferred for the three sizes (the full set of changes), and before vs main.

  • Computed census, 923 cells (font size, weight, line height, letter spacing, radius, transition duration and easing, z-index): before vs deferred: radius, transition duration, easing and z-index differ in 0 cells; 8 cells differ by 1-2 elements, in the mock terminal's text. Before vs main (changes nothing in these screens) shows the same 8 cells, so that is run-to-run noise.
  • Pixels (channel diff > 8/255): main vs before 0.00-0.13 % in 34 images (1.07 % and 95 % on the two ink sign-in images, a backdrop fade caught at a different point); deferred vs before 0.00-0.22 % in 35 images and the same 95 % sign-in outlier. Same band as unchanged code.
  • The one intended change: shared components get their transition inside the app (probe table in the previous comment; on the deferred branch every screen now reports 0.14s | cubic-bezier(0.23, 1, 0.32, 1) for .check).

Tables: diff-before3-vs-after3.tsv, diff-before3-vs-after-deferred3.tsv (per image). Full-resolution screenshots (DPR 2 and 3) are in ~claude/.cache/codex-scratch/j195/evidence/ui-audit/{before3,after3,after-deferred3} on compute01.

**Contact sheets and regression proof (J195)** — production build, 36 captures per phase (6 screens x 3 sizes x ink/paper). Phases: `before` = origin/dev a89ebb0ae + spec (941569815), `main` = `ui-consistency-j195`, `deferred` = `ui-consistency-j195-deferred` (7 more commits, not for merge yet). Each sheet: rows canvas, sidebar, palette, settings, popover, sign-in; columns ink before, ink after, paper before, paper after. Attached below: before vs deferred for the three sizes (the full set of changes), and before vs main. - Computed census, 923 cells (font size, weight, line height, letter spacing, radius, transition duration and easing, z-index): before vs deferred: radius, transition duration, easing and z-index differ in 0 cells; 8 cells differ by 1-2 elements, in the mock terminal's text. Before vs main (changes nothing in these screens) shows the same 8 cells, so that is run-to-run noise. - Pixels (channel diff > 8/255): main vs before 0.00-0.13 % in 34 images (1.07 % and 95 % on the two ink sign-in images, a backdrop fade caught at a different point); deferred vs before 0.00-0.22 % in 35 images and the same 95 % sign-in outlier. Same band as unchanged code. - The one intended change: shared components get their transition inside the app (probe table in the previous comment; on the deferred branch every screen now reports `0.14s | cubic-bezier(0.23, 1, 0.32, 1)` for `.check`). Tables: `diff-before3-vs-after3.tsv`, `diff-before3-vs-after-deferred3.tsv` (per image). Full-resolution screenshots (DPR 2 and 3) are in `~claude/.cache/codex-scratch/j195/evidence/ui-audit/{before3,after3,after-deferred3}` on compute01.
sheet-393x852-before3-vs-after-deferred3.png
224 KiB
sheet-375x667-before3-vs-after-deferred3.png
268 KiB
sheet-1440x900-before3-vs-after3.png
283 KiB
sheet-393x852-before3-vs-after3.png
225 KiB
sheet-375x667-before3-vs-after3.png
262 KiB
sheet-1440x900-before3-vs-after-deferred3.png
278 KiB
sheet-393x852-before3-vs-after-deferred3.png sheet-375x667-before3-vs-after-deferred3.png sheet-1440x900-before3-vs-after3.png sheet-393x852-before3-vs-after3.png sheet-375x667-before3-vs-after3.png sheet-1440x900-before3-vs-after-deferred3.png
kayg commented 2026-11-04 12:46:15 +00:00
Author
Owner
Copy link

Gate, first run (early signal; head was d3ca87fb3, a SHA that no longer exists on the branch: the history was rewritten afterwards to fold the spec fixes and the doc into their commits. Rust and the tracer sources are unchanged since; the final-head run is queued).

quick scope vs merge base with origin/dev: workspace=0 packages=[] tracer=1
| fmt | PASS | ui literals | PASS | tracer typecheck | PASS |        QUICK_EXIT=0

| Rust gate                 | Result       |
| fmt                       | PASS         |
| VT native fmt             | PASS         |
| CLI plugin compile-out    | PASS         |
| clippy                    | PASS         |
| SSH runtime test helper   | PASS         |
| real-binary fixture paths | PASS         |
| Traefik component manifest | PASS         |
| release component publication | PASS         |
| test                      | FAIL (101)   |
| SSH inode reuse fixture   | PASS         |
| shim process-per-user     | PASS         |
| cargo-deny                | PASS         |
| gate architecture         | PASS         |
| static builds             | PASS         |
VERIFY_EXIT=1

The test row fails on 7 tests in crates/infraternal, all in restart (restart::tests::{reload_reopens_install_path_when_the_running_inode_was_replaced, rootless_even_marked_replacement_is_cold_and_open_fd_pins_the_inode, cold_exec_with_closed_stdin_keeps_the_image_pinned_and_maps_null, real_fd_exec_pins_replacement_and_cold_rollback_uses_null_stdin}, restart::recovery_tests::verified_exec_and_late_refusal_preserve_listener_and_ca, restart_binary::audit_tests::{pathname_replacement_keeps_the_admitted_inode, admitted_inode_cannot_change_in_place_before_execution}), each Err(Kind(PermissionDenied)) from Signed::...open() as the non-root claude user. This branch changes no Rust (git diff origin/dev..HEAD -- crates Cargo.toml Cargo.lock is empty), and the same tests appear failing in older logs on this VM (~/d315-fast.log, ~/flake-verify.log). The final-head run also runs them on the unmodified base a89ebb0ae to prove it.

Deferred branch verify quick: PASS (ui literals added vs merge base: font-size-px -1, radius-px -27, ease-curve -46).

**Gate, first run (early signal; head was d3ca87fb3, a SHA that no longer exists on the branch: the history was rewritten afterwards to fold the spec fixes and the doc into their commits. Rust and the tracer sources are unchanged since; the final-head run is queued).** ``` quick scope vs merge base with origin/dev: workspace=0 packages=[] tracer=1 | fmt | PASS | ui literals | PASS | tracer typecheck | PASS | QUICK_EXIT=0 | Rust gate | Result | | fmt | PASS | | VT native fmt | PASS | | CLI plugin compile-out | PASS | | clippy | PASS | | SSH runtime test helper | PASS | | real-binary fixture paths | PASS | | Traefik component manifest | PASS | | release component publication | PASS | | test | FAIL (101) | | SSH inode reuse fixture | PASS | | shim process-per-user | PASS | | cargo-deny | PASS | | gate architecture | PASS | | static builds | PASS | VERIFY_EXIT=1 ``` The `test` row fails on 7 tests in `crates/infraternal`, all in `restart` (`restart::tests::{reload_reopens_install_path_when_the_running_inode_was_replaced, rootless_even_marked_replacement_is_cold_and_open_fd_pins_the_inode, cold_exec_with_closed_stdin_keeps_the_image_pinned_and_maps_null, real_fd_exec_pins_replacement_and_cold_rollback_uses_null_stdin}`, `restart::recovery_tests::verified_exec_and_late_refusal_preserve_listener_and_ca`, `restart_binary::audit_tests::{pathname_replacement_keeps_the_admitted_inode, admitted_inode_cannot_change_in_place_before_execution}`), each `Err(Kind(PermissionDenied))` from `Signed::...open()` as the non-root `claude` user. This branch changes no Rust (`git diff origin/dev..HEAD -- crates Cargo.toml Cargo.lock` is empty), and the same tests appear failing in older logs on this VM (`~/d315-fast.log`, `~/flake-verify.log`). The final-head run also runs them on the unmodified base a89ebb0ae to prove it. Deferred branch `verify quick`: PASS (`ui literals added vs merge base: font-size-px -1, radius-px -27, ease-curve -46`).
kayg commented 2026-11-04 13:42:13 +00:00
Author
Owner
Copy link

Finish (J195) — ui-consistency-j195 head 1c6a1562e08948890e86ad148cbc7de0daa06626 (6 commits on origin/dev a89ebb0ae), stacked ui-consistency-j195-deferred head 7474214f084f0f17d0643a1905d84d1ea270bc30 (7 commits more). Nothing merged, nothing deployed.

Audit document: docs/design/UI-CONSISTENCY-AUDIT.md on the main branch (register, counts, deliberate differences with their decision, the ten owner questions, the deferred files).

Counts (3390 style declarations scanned in apps/tracer + packages/ui): (a) accidental: 122 declarations (radius 55 on a scale step or 999px, easing 57, letter spacing 6, and 4 lengths in doc.css); fixed on the main branch 8 (4 pasted curves and 4 lengths in doc.css, free files), ready on the deferred branch 114 (55 radius, 39 pasted curves, 14 reads of the undefined --ease, 6 letter spacings; pixel-identical except the --ease fix). (b) deliberate: 54 radii + 22 resets + 10 font sizes + 63 hairline widths + 41 ring shadows + touch targets, each documented in the audit (the code comment is added where the file was free: copy-feedback.ts, doc.css). (c) unclear, left alone: 65 radii, 70 easings, 23 tracking values, 17 weights, 49 font sizes, 572 off-grid lengths, 85 icon sizes, 83 z-indexes, 9 durations, 4 height tokens: 10 owner questions. Deferred: 2939 of 3390 declarations (87 %) are in files an open branch touches.

Guard: verify quick row "ui literals" (0 s): a change may not add px font sizes, px radii over 2 px or pasted cubic-bezier(0.23,1,0.32,1); opt out with ui-literal: <decision>.

Proven in a production build: shared .check and .swap (copy-tick morph, D103) have no transition inside the app (0s | ease) and the sign-in card has (0.14s | cubic-bezier(0.23, 1, 0.32, 1)); fixed on the deferred branch. Contact sheets and the 923-cell census comparison are in the previous comments.

D188 (JavaScript transfer budget), bun run transfer-budget.ts, caps unchanged (critical 327680 gzip / 273408 brotli):

head critical identity critical gzip critical brotli deferred gzip
base a89ebb0ae 1024570 318890 266223 54328
main 1c6a1562e 1024570 318898 266454 54340
deferred 7474214f0 1024553 318909 266445 54333

JavaScript content did not grow (identity: 0 / −17 B). The compressed sizes move by 8–231 B between builds of identical JavaScript (main has no JavaScript change), so that is build-to-build variation, not growth.

Gate output, verbatim, on head 1c6a1562e (compute01, scripts/verify quick then scripts/verify fast):

GATE HEAD 1c6a1562e08948890e86ad148cbc7de0daa06626 base a89ebb0ae
quick scope vs merge base with origin/dev: workspace=0 packages=[] tracer=1

==> fmt
(fmt: 5s)

==> ui literals
ui literals added vs merge base: font-size-px -1, radius-px -1, ease-curve -4
(ui literals: 0s)

==> tracer typecheck
bun install v1.4.2 (744846f84)

Checked 38 installs across 69 packages (no changes) [26.00ms]
$ bun x tsc --project tsconfig.json
(tracer typecheck: 15s)

+---------------------------+--------------+
| Rust gate                 | Result       |
+---------------------------+--------------+
| fmt                       | PASS         |
| ui literals               | PASS         |
| tracer typecheck          | PASS         |
+---------------------------+--------------+

QUICK pre-check passed. This is not the gate: run scripts/verify fast before reporting done.
QUICK_EXIT=0

==> Repository artifact size gate
repository artifact gate: PASS (8 changed path(s), merge base a89ebb0ae71d)

==> Release deployment functional checks
.....Release build failed for aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa; gate log retained
.Build deferred: host capacity limit reached
.........
----------------------------------------------------------------------
Ran 15 tests in 0.652s

[... verify fast ...]
| Rust gate                 | Result       |
PT_INTERP: absent; size: 23373616 bytes; baseline 22418256; delta +955360 bytes (report only, D243 amended 2026-11-01)
PT_INTERP: absent; size: 21244048 bytes; baseline 20453280; delta +790768 bytes (report only, D243 amended 2026-11-01)
x86_64-unknown-linux-musl: +102400 bytes accepted with delta; zero/negative deltas reported; missing/wrong/dynamic/PT_INTERP/inspection defects rejected PASS
aarch64-unknown-linux-musl: +102400 bytes accepted with delta; zero/negative deltas reported; missing/wrong/dynamic/PT_INTERP/inspection defects rejected PASS

[... TypeScript half: ts_verify_fast (verify fast stops at the failing Rust row before reaching it) ...]
 0 fail
 77248 expect() calls
Ran 48 tests across 12 files. [14.91s]
TS_VERIFY_EXIT=0

verify fast fails on its test row, which is a failure that also happens on unmodified origin/dev: 7 tests in crates/infraternal (restart::tests::{reload_reopens_install_path_when_the_running_inode_was_replaced, rootless_even_marked_replacement_is_cold_and_open_fd_pins_the_inode, cold_exec_with_closed_stdin_keeps_the_image_pinned_and_maps_null, real_fd_exec_pins_replacement_and_cold_rollback_uses_null_stdin}, restart::recovery_tests::verified_exec_and_late_refusal_preserve_listener_and_ca, restart_binary::audit_tests::{pathname_replacement_keeps_the_admitted_inode, admitted_inode_cannot_change_in_place_before_execution}), each Err(Kind(PermissionDenied)) for the non-root claude user. Proof on unmodified a89ebb0ae, same VM and cache: cargo test --locked -p infraternal --bins restart -- --test-threads=1 gives test result: FAILED. 21 passed; 7 failed; 1 ignored with the same seven names. This branch has no Rust or manifest difference (git diff a89ebb0ae 1c6a1562e -- crates Cargo.toml Cargo.lock is empty). Because the script stops at that row, ts_verify_fast (where D188 lives) did not run inside fast; I ran it on its own on the same head: TS_VERIFY_EXIT=0. Static builds row PASS; its report-only delta (+955360 x86, +790768 arm vs the baseline file) is not from this branch (no Rust changed).

Deferred branch verify quick on 7474214f0: PASS (ui literals added vs merge base: font-size-px -1, radius-px -27, ease-curve -46; fmt, tracer typecheck PASS).

Owner questions (details in section 6 of the audit): 1 two type systems and the weights 550/650 (touch text scale vs D174); 2 radius steps between the scale (four radii for floating cards: 24.2 / 19.8 / 17.6 / 15.4 px); 3 the 4 px grid vs 6/10/14 px half steps; 4 four near-equal row/control height tokens; 5 icon sizes vs --ui-icon; 6 a z-index scale; 7 duration steps (--t-pop/--t-panel unused); 8 a TypeScript EASE_OUT constant; 9 standalone docs type is 13 px while the app is about 14.3 px; 10 title sizes. Recommendation for each is in the audit.

Deferred (waiting on j129 / j132 / j193 / j166 / sidebar-292 / toast-stack-j165): ui-consistency-j195-deferred. After they merge, rebase it (each commit is one mechanical pass, described in its message), re-run UI_AUDIT=1 UI_AUDIT_LABEL=after bun x playwright test ui-consistency-audit, compare with the sheets here.

Cleanup: target dir (46 GB) and /tmp/j195s removed; VM worktrees ~/worktrees/j195, ~/worktrees/j195g and the evidence (~/.cache/codex-scratch/j195/evidence/ui-audit, 34 MB) kept.

**Finish (J195) — `ui-consistency-j195` head `1c6a1562e08948890e86ad148cbc7de0daa06626` (6 commits on origin/dev a89ebb0ae), stacked `ui-consistency-j195-deferred` head `7474214f084f0f17d0643a1905d84d1ea270bc30` (7 commits more). Nothing merged, nothing deployed.** Audit document: `docs/design/UI-CONSISTENCY-AUDIT.md` on the main branch (register, counts, deliberate differences with their decision, the ten owner questions, the deferred files). **Counts (3390 style declarations scanned in `apps/tracer` + `packages/ui`):** (a) accidental: 122 declarations (radius 55 on a scale step or `999px`, easing 57, letter spacing 6, and 4 lengths in `doc.css`); fixed on the main branch 8 (4 pasted curves and 4 lengths in `doc.css`, free files), ready on the deferred branch 114 (55 radius, 39 pasted curves, 14 reads of the undefined `--ease`, 6 letter spacings; pixel-identical except the `--ease` fix). (b) deliberate: 54 radii + 22 resets + 10 font sizes + 63 hairline widths + 41 ring shadows + touch targets, each documented in the audit (the code comment is added where the file was free: `copy-feedback.ts`, `doc.css`). (c) unclear, left alone: 65 radii, 70 easings, 23 tracking values, 17 weights, 49 font sizes, 572 off-grid lengths, 85 icon sizes, 83 z-indexes, 9 durations, 4 height tokens: 10 owner questions. Deferred: 2939 of 3390 declarations (87 %) are in files an open branch touches. **Guard:** `verify quick` row "ui literals" (0 s): a change may not add px font sizes, px radii over 2 px or pasted `cubic-bezier(0.23,1,0.32,1)`; opt out with `ui-literal: <decision>`. **Proven in a production build:** shared `.check` and `.swap` (copy-tick morph, D103) have no transition inside the app (`0s | ease`) and the sign-in card has (`0.14s | cubic-bezier(0.23, 1, 0.32, 1)`); fixed on the deferred branch. Contact sheets and the 923-cell census comparison are in the previous comments. **D188 (JavaScript transfer budget), `bun run transfer-budget.ts`, caps unchanged (critical 327680 gzip / 273408 brotli):** | head | critical identity | critical gzip | critical brotli | deferred gzip | |---|---|---|---|---| | base a89ebb0ae | 1024570 | 318890 | 266223 | 54328 | | main 1c6a1562e | 1024570 | 318898 | 266454 | 54340 | | deferred 7474214f0 | 1024553 | 318909 | 266445 | 54333 | JavaScript content did not grow (identity: 0 / −17 B). The compressed sizes move by 8–231 B between builds of identical JavaScript (main has no JavaScript change), so that is build-to-build variation, not growth. **Gate output, verbatim, on head 1c6a1562e (compute01, `scripts/verify quick` then `scripts/verify fast`):** ``` GATE HEAD 1c6a1562e08948890e86ad148cbc7de0daa06626 base a89ebb0ae quick scope vs merge base with origin/dev: workspace=0 packages=[] tracer=1 ==> fmt (fmt: 5s) ==> ui literals ui literals added vs merge base: font-size-px -1, radius-px -1, ease-curve -4 (ui literals: 0s) ==> tracer typecheck bun install v1.4.2 (744846f84) Checked 38 installs across 69 packages (no changes) [26.00ms] $ bun x tsc --project tsconfig.json (tracer typecheck: 15s) +---------------------------+--------------+ | Rust gate | Result | +---------------------------+--------------+ | fmt | PASS | | ui literals | PASS | | tracer typecheck | PASS | +---------------------------+--------------+ QUICK pre-check passed. This is not the gate: run scripts/verify fast before reporting done. QUICK_EXIT=0 ==> Repository artifact size gate repository artifact gate: PASS (8 changed path(s), merge base a89ebb0ae71d) ==> Release deployment functional checks .....Release build failed for aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa; gate log retained .Build deferred: host capacity limit reached ......... ---------------------------------------------------------------------- Ran 15 tests in 0.652s [... verify fast ...] | Rust gate | Result | PT_INTERP: absent; size: 23373616 bytes; baseline 22418256; delta +955360 bytes (report only, D243 amended 2026-11-01) PT_INTERP: absent; size: 21244048 bytes; baseline 20453280; delta +790768 bytes (report only, D243 amended 2026-11-01) x86_64-unknown-linux-musl: +102400 bytes accepted with delta; zero/negative deltas reported; missing/wrong/dynamic/PT_INTERP/inspection defects rejected PASS aarch64-unknown-linux-musl: +102400 bytes accepted with delta; zero/negative deltas reported; missing/wrong/dynamic/PT_INTERP/inspection defects rejected PASS [... TypeScript half: ts_verify_fast (verify fast stops at the failing Rust row before reaching it) ...] 0 fail 77248 expect() calls Ran 48 tests across 12 files. [14.91s] TS_VERIFY_EXIT=0 ``` `verify fast` fails on its `test` row, which is a failure that also happens on unmodified origin/dev: 7 tests in `crates/infraternal` (`restart::tests::{reload_reopens_install_path_when_the_running_inode_was_replaced, rootless_even_marked_replacement_is_cold_and_open_fd_pins_the_inode, cold_exec_with_closed_stdin_keeps_the_image_pinned_and_maps_null, real_fd_exec_pins_replacement_and_cold_rollback_uses_null_stdin}`, `restart::recovery_tests::verified_exec_and_late_refusal_preserve_listener_and_ca`, `restart_binary::audit_tests::{pathname_replacement_keeps_the_admitted_inode, admitted_inode_cannot_change_in_place_before_execution}`), each `Err(Kind(PermissionDenied))` for the non-root `claude` user. Proof on unmodified a89ebb0ae, same VM and cache: `cargo test --locked -p infraternal --bins restart -- --test-threads=1` gives `test result: FAILED. 21 passed; 7 failed; 1 ignored` with the same seven names. This branch has no Rust or manifest difference (`git diff a89ebb0ae 1c6a1562e -- crates Cargo.toml Cargo.lock` is empty). Because the script stops at that row, `ts_verify_fast` (where D188 lives) did not run inside `fast`; I ran it on its own on the same head: `TS_VERIFY_EXIT=0`. Static builds row PASS; its report-only delta (`+955360` x86, `+790768` arm vs the baseline file) is not from this branch (no Rust changed). Deferred branch `verify quick` on 7474214f0: PASS (`ui literals added vs merge base: font-size-px -1, radius-px -27, ease-curve -46`; fmt, tracer typecheck PASS). **Owner questions** (details in section 6 of the audit): 1 two type systems and the weights 550/650 (touch text scale vs D174); 2 radius steps between the scale (four radii for floating cards: 24.2 / 19.8 / 17.6 / 15.4 px); 3 the 4 px grid vs 6/10/14 px half steps; 4 four near-equal row/control height tokens; 5 icon sizes vs `--ui-icon`; 6 a z-index scale; 7 duration steps (`--t-pop`/`--t-panel` unused); 8 a TypeScript `EASE_OUT` constant; 9 standalone docs type is 13 px while the app is about 14.3 px; 10 title sizes. Recommendation for each is in the audit. **Deferred (waiting on j129 / j132 / j193 / j166 / sidebar-292 / toast-stack-j165):** `ui-consistency-j195-deferred`. After they merge, rebase it (each commit is one mechanical pass, described in its message), re-run `UI_AUDIT=1 UI_AUDIT_LABEL=after bun x playwright test ui-consistency-audit`, compare with the sheets here. Cleanup: target dir (46 GB) and `/tmp/j195s` removed; VM worktrees `~/worktrees/j195`, `~/worktrees/j195g` and the evidence (`~/.cache/codex-scratch/j195/evidence/ui-audit`, 34 MB) kept.
kayg commented 2026-11-08 18:40:09 +00:00
Author
Owner
Copy link

Owner answers, 2026-11-08 (inline grill), recorded as D372.

The owner chose "Merge fixes, ask the 10". The questions were answered as recommended:

  1. One type system, with the phone size set once.
  2. Add --r-xs, keep nested radii, and round the rest.
  3. Allow 2 px half steps on the grid.
  4. Merge the control and row heights.
  5. Add --ui-icon-sm.
  6. Three durations: 120, 180 and 240 ms.
  7. The docs pages follow the app's root size.
  8. One title role, at 22 px.

The z-index layer scale and the EASE_OUT export are decided as in-scope engineering.

**Owner answers, 2026-11-08 (inline grill), recorded as D372.** The owner chose "Merge fixes, ask the 10". The questions were answered as recommended: 1. One type system, with the phone size set once. 2. Add `--r-xs`, keep nested radii, and round the rest. 3. Allow 2 px half steps on the grid. 4. Merge the control and row heights. 5. Add `--ui-icon-sm`. 6. Three durations: 120, 180 and 240 ms. 7. The docs pages follow the app's root size. 8. One title role, at 22 px. The z-index layer scale and the `EASE_OUT` export are decided as in-scope engineering.
kayg commented 2026-11-08 18:43:39 +00:00
Author
Owner
Copy link

J383 start (2026-11-08). Part 1 branch ui-consistency-j195-deferred-rebased: the 7 commits of ui-consistency-j195-deferred (7474214f0) rebased on origin/dev a894dd877. One conflict (toast transition, packages/ui/styles.css), resolved for dev's newer toast behaviour (touch-action, translate/scale/height transitions) with the --ease-out read kept. Residual scan after the rebase: 0 pasted curves in CSS, 0 var(--ease) reads outside the two files that define it, 0 999px, 0 -0.01em (the remaining curve literals are the TypeScript Web Animations ones, part 2). Next: before/after production-build captures (1440×900, 393×852, 375×667; ink and paper) on dev-02, then verify quick.
Part 2 branch ui-consistency-d372 will be stacked on it. Plan and estimate (~6 h): type roles, --r-xs, grid, heights, icons, durations, docs, title role, z-index scale, EASE_OUT; one atomic commit per item, each with contact sheets (phone first).

**J383 start (2026-11-08).** Part 1 branch `ui-consistency-j195-deferred-rebased`: the 7 commits of `ui-consistency-j195-deferred` (7474214f0) rebased on origin/dev a894dd877. One conflict (toast `transition`, `packages/ui/styles.css`), resolved for dev's newer toast behaviour (touch-action, translate/scale/height transitions) with the `--ease-out` read kept. Residual scan after the rebase: 0 pasted curves in CSS, 0 `var(--ease)` reads outside the two files that define it, 0 `999px`, 0 `-0.01em` (the remaining curve literals are the TypeScript Web Animations ones, part 2). Next: before/after production-build captures (1440×900, 393×852, 375×667; ink and paper) on dev-02, then `verify quick`. Part 2 branch `ui-consistency-d372` will be stacked on it. Plan and estimate (~6 h): type roles, `--r-xs`, grid, heights, icons, durations, docs, title role, z-index scale, `EASE_OUT`; one atomic commit per item, each with contact sheets (phone first).
kayg commented 2026-11-08 19:25:30 +00:00
Author
Owner
Copy link

J383 part 1 evidence: ui-consistency-j195-deferred-rebased, head b2f496177cf7af61f8be7ba59ccb85310ad46d79 (7 commits on origin/dev a894dd877, pushed).

Production-build captures, 6 screens × 3 sizes × ink/paper (32 phone+desktop images per phase, Settings is not captured on phones), before = origin/dev a894dd877, after = this branch. Evidence on dev-02: ~/j383-evidence/ui-audit/p1-before and p1-after.

  • Computed census (font size, weight, line height, letter spacing, radius, transition duration and easing, z-index): 0 cells differ between before and after. The only difference is the 26 probe entries: .check goes from 0s | ease to 0.14s | cubic-bezier(0.23, 1, 0.32, 1) and .swap svg from 0s | ease to 0.15s | cubic-bezier(...). That is the --ease fix, the expected change.
  • Pixels (channel delta > 8/255): the worst image is 1440×900 paper sidebar with 3197 changed pixels (0.25 %), then 945, 797 and 554. The diff image of the worst one is a regular dot pattern, the canvas dot grid, not an element this branch touches. I did not run a before-vs-before control, so "capture noise" is my reading, not a measurement. The J195 run saw the same band (0.00–0.22 %).
  • The rebase had one conflict, the toast transition in packages/ui/styles.css. Dev's newer rule wins (touch-action: pan-x plus translate, scale and height transitions); only the var(--ease-out) read is kept. J304 and #427 did not conflict. After the rebase there are 0 pasted curves in CSS, 0 reads of the undefined --ease outside auth.css and docs.css (they define it), 0 999px radii and 0 -0.01em.
  • verify quick is queued on dev-02 (position 3); I will post it verbatim.
**J383 part 1 evidence: `ui-consistency-j195-deferred-rebased`, head `b2f496177cf7af61f8be7ba59ccb85310ad46d79` (7 commits on origin/dev a894dd877, pushed).** Production-build captures, 6 screens × 3 sizes × ink/paper (32 phone+desktop images per phase, Settings is not captured on phones), before = origin/dev a894dd877, after = this branch. Evidence on dev-02: `~/j383-evidence/ui-audit/p1-before` and `p1-after`. - Computed census (font size, weight, line height, letter spacing, radius, transition duration and easing, z-index): **0 cells differ** between before and after. The only difference is the 26 probe entries: `.check` goes from `0s | ease` to `0.14s | cubic-bezier(0.23, 1, 0.32, 1)` and `.swap svg` from `0s | ease` to `0.15s | cubic-bezier(...)`. That is the `--ease` fix, the expected change. - Pixels (channel delta > 8/255): the worst image is 1440×900 paper sidebar with 3197 changed pixels (0.25 %), then 945, 797 and 554. The diff image of the worst one is a regular dot pattern, the canvas dot grid, not an element this branch touches. I did not run a before-vs-before control, so "capture noise" is my reading, not a measurement. The J195 run saw the same band (0.00–0.22 %). - The rebase had one conflict, the toast `transition` in `packages/ui/styles.css`. Dev's newer rule wins (`touch-action: pan-x` plus translate, scale and height transitions); only the `var(--ease-out)` read is kept. J304 and #427 did not conflict. After the rebase there are 0 pasted curves in CSS, 0 reads of the undefined `--ease` outside `auth.css` and `docs.css` (they define it), 0 `999px` radii and 0 `-0.01em`. - `verify quick` is queued on dev-02 (position 3); I will post it verbatim.
p1-1440x900-before-after.png
1.8 MiB
p1-375x667-before-after.png
2.8 MiB
p1-393x852-before-after.png
3.2 MiB
p1-1440x900-before-after.png p1-375x667-before-after.png p1-393x852-before-after.png
kayg commented 2026-11-08 19:43:21 +00:00
Author
Owner
Copy link

J383 progress (2026-11-08): part 2 branch ui-consistency-d372 pushed (stacked on part 1 head c27a7ad2d). One atomic commit per D372 item: durations, EASE_OUT, --r-xs and radius rounding, one control height, --ui-icon-sm, grid, type roles, title role, docs.css, six-step layer scale, overlay-pair spec, TASTE. Captures and gates are queued on dev-02 (the slot has 7 jobs ahead); results follow.

Judgement calls for the orchestrator (I proceeded with the default; each is one commit to change):

  1. Type, phone sizes. D372 says "body 15 px, labels 13 px on phones". The roles are 0.875 rem body and 0.8125 rem labels, and the sign-in card renders them as 15.4 and 14.3 px at the 110 % root. I read the numbers as unscaled rem, the old touch values: body 0.9375 rem on touch (set once, --ui-size-body in the touch block), 0.875 rem on desktop, labels 0.8125 rem everywhere. Old --text-* names remain as aliases of the role sizes (body-in-row = label, secondary and caption = eyebrow); --text-input stays 16 px on touch for iOS. The visible effect on a phone: the old touch text (16/15/13 px) shrinks to 13/12/12. If the owner meant rendered pixels, only the touch block changes.
  2. Title size. 22 px at the default 110 % root is 1.25 rem; a phone has a 16 px root, so a title is 20 px there. 1.375 rem would give 22 px on a phone and 24.2 px on the desktop default.
  3. Radii. Off-scale values round to the nearest step, ties to the smaller. The three 1.75 rem frames (cluster and mesh) have no step above --r-lg (1.125 rem), so I left them alone rather than shrink them by 10 px. Say if they should round to --r-lg.
  4. Heights. --ui-row-h is deleted and its 14 reads read --ui-control-h (2 rem). Controls grow from 30 to 32 px.
  5. Layers. The scale is monotone from the old numbers (see the commit), so no pair changes order; inside a step the old rank is a calc() offset. 0 and -1 stay.
  6. Durations. Only transition declarations moved (110–150 ms to 120, 160–200 to 180, 220–260 to 240); animations, keyframes and delays were left.
**J383 progress (2026-11-08): part 2 branch `ui-consistency-d372` pushed (stacked on part 1 head `c27a7ad2d`).** One atomic commit per D372 item: durations, `EASE_OUT`, `--r-xs` and radius rounding, one control height, `--ui-icon-sm`, grid, type roles, title role, docs.css, six-step layer scale, overlay-pair spec, TASTE. Captures and gates are queued on dev-02 (the slot has 7 jobs ahead); results follow. Judgement calls for the orchestrator (I proceeded with the default; each is one commit to change): 1. **Type, phone sizes.** D372 says "body 15 px, labels 13 px on phones". The roles are 0.875 rem body and 0.8125 rem labels, and the sign-in card renders them as 15.4 and 14.3 px at the 110 % root. I read the numbers as unscaled rem, the old touch values: body 0.9375 rem on touch (set once, `--ui-size-body` in the touch block), 0.875 rem on desktop, labels 0.8125 rem everywhere. Old `--text-*` names remain as aliases of the role sizes (body-in-row = label, secondary and caption = eyebrow); `--text-input` stays 16 px on touch for iOS. The visible effect on a phone: the old touch text (16/15/13 px) shrinks to 13/12/12. If the owner meant rendered pixels, only the touch block changes. 2. **Title size.** 22 px at the default 110 % root is 1.25 rem; a phone has a 16 px root, so a title is 20 px there. 1.375 rem would give 22 px on a phone and 24.2 px on the desktop default. 3. **Radii.** Off-scale values round to the nearest step, ties to the smaller. The three 1.75 rem frames (cluster and mesh) have no step above `--r-lg` (1.125 rem), so I left them alone rather than shrink them by 10 px. Say if they should round to `--r-lg`. 4. **Heights.** `--ui-row-h` is deleted and its 14 reads read `--ui-control-h` (2 rem). Controls grow from 30 to 32 px. 5. **Layers.** The scale is monotone from the old numbers (see the commit), so no pair changes order; inside a step the old rank is a `calc()` offset. 0 and -1 stay. 6. **Durations.** Only `transition` declarations moved (110–150 ms to 120, 160–200 to 180, 220–260 to 240); animations, keyframes and delays were left.
kayg commented 2026-11-08 22:57:49 +00:00
Author
Owner
Copy link

J383 part 1 finish: ui-consistency-j195-deferred-rebased, head c27a7ad2de8d196b28d85befca777d4bf3d11dbf (8 commits on origin/dev a894dd877: the 7 of J195 plus a doc comment on the status-line stylesheet constant, which the D345 gate asked for). Captures and census are in the comment above.

scripts/verify quick on that head, verbatim:


+---------------------------+--------------+
| Rust gate                 | Result       |
+---------------------------+--------------+
| scope guard               | PASS         |
| migration names           | PASS         |
| Nix Cargo vendoring       | PASS         |
| fmt                       | PASS         |
| doc coverage ratchet      | PASS         |
| shape lints off (D351)    | PASS         |
| doc comments on diff      | PASS         |
| fixed test waits          | PASS         |
| ui literals               | PASS         |
| tracer typecheck          | PASS         |
+---------------------------+--------------+

QUICK pre-check passed. This is not the gate: run scripts/verify fast before reporting done.
QUICK_EXIT=0

Not run: scripts/verify fast (quick is the pre-check; the merge gate runs on compute01). Browser specs on a production build: the 36-image audit capture above is the production-build run; the three feature specs run in the baseline phase of the part 2 job, results follow there.

**J383 part 1 finish: `ui-consistency-j195-deferred-rebased`, head `c27a7ad2de8d196b28d85befca777d4bf3d11dbf`** (8 commits on origin/dev a894dd877: the 7 of J195 plus a doc comment on the status-line stylesheet constant, which the D345 gate asked for). Captures and census are in the comment above. `scripts/verify quick` on that head, verbatim: ``` +---------------------------+--------------+ | Rust gate | Result | +---------------------------+--------------+ | scope guard | PASS | | migration names | PASS | | Nix Cargo vendoring | PASS | | fmt | PASS | | doc coverage ratchet | PASS | | shape lints off (D351) | PASS | | doc comments on diff | PASS | | fixed test waits | PASS | | ui literals | PASS | | tracer typecheck | PASS | +---------------------------+--------------+ QUICK pre-check passed. This is not the gate: run scripts/verify fast before reporting done. QUICK_EXIT=0 ``` Not run: `scripts/verify fast` (quick is the pre-check; the merge gate runs on compute01). Browser specs on a production build: the 36-image audit capture above is the production-build run; the three feature specs run in the baseline phase of the part 2 job, results follow there.
kayg commented 2026-11-09 00:29:14 +00:00
Author
Owner
Copy link

J383 part 2 finish: ui-consistency-d372, head af1dc7db324ba2c9b5cd29cc508c0c5ec30f6521 (13 commits stacked on part 1 head c27a7ad2d). Nothing merged, nothing deployed.

Commits: durations (3 steps), EASE_OUT, --r-xs and radius rounding, one control height, --ui-icon-sm, grid (3/5 px round to 4), type roles, one title role, docs.css, six-step layer scale, overlay-pair spec, TASTE, doc comments (D345).

scripts/verify quick on the head, verbatim:

+---------------------------+--------------+
| Rust gate                 | Result       |
+---------------------------+--------------+
| scope guard               | PASS         |
| migration names           | PASS         |
| Nix Cargo vendoring       | PASS         |
| fmt                       | PASS         |
| doc coverage ratchet      | PASS         |
| shape lints off (D351)    | PASS         |
| doc comments on diff      | PASS         |
| fixed test waits          | PASS         |
| ui literals               | PASS         |
| tracer typecheck          | PASS         |
+---------------------------+--------------+

QUICK pre-check passed. This is not the gate: run scripts/verify fast before reporting done.
QUICK_EXIT=0

Bundle budget (bun run build.ts && bun run transfer-budget.ts, exit 0, no cap touched): critical gzip 319,018 / 327,680 (part 1: 319,007), deferred gzip 58,560 / 61,440 (58,546), deferred brotli 51,916 / 55,296 (51,933). Log: ~/j383-p2b/j383-p2-budget.log on dev-02.

Census (production build, 1440×900, 393×852, 375×667, ink and paper, 6 screens), part 1 head vs part 2:

  • transition durations: 9 values (0.12 to 0.26 s) become 3 (0.12 s ×1398, 0.18 s ×480, 0.24 s ×56);
  • z-index: 5→20, 6→30, 4→10, 7→31, 8→32, 12→35, and 0, 1, 2, 3, −1 unchanged (monotone, as designed);
  • radii and font sizes: the old touch body (16 px), secondary (15 px) and caption (13 px) are gone; the off-scale radii (5.5, 9, 10, 15, 16, 24.2 px etc.) are gone.

Overlay pairs (ui-layer-pairs.spec.ts, hit test with elementFromPoint): before and after both 28 of 30 pairs have the upper layer on top, with the same result in every cell. The 2 that fail in both runs are tooltip-over-sidebar at 1440×900 (ink and paper): my driver could not make the tooltip appear, so that pair is not proven by capture (the tooltip layers map 1000/1001/1002 to --z-tooltip + 0/1/2). Phones skip Settings and hover-tooltip pairs (Settings is not reachable on a phone in the mock).

Browser specs on a production build (dev-02): docs, docs-reading, docs-palette, icon-text-centre, terminal-links, toast-stack, tooltip-fit, settings-rows, settings-edges, sidebar-polish, palette-polish, palette-navigation, row-menu, addhost-wizard, replacement-alignment, lock-icons: 165 passed in the first run. 13 failed; the rerun after my fixes:

  • Fixed (specs that pinned old px values; the new values follow D372): docs-reading and docs (desktop: 17 px icon → 18.7, 1.5rem and 16 px offsets are now root relative), replacement-alignment (a touch rename field stays 16 px for iOS, above the 13 px label, so the spec allows max(label, 16) on touch). Reruns: docs and docs-reading 9 passed; replacement-alignment and icon-text-centre green in the combined rerun (log ~/j383-c/p2-specs.log).
  • Pre-existing, identical on part 1 head (and so not caused by D372): docs-palette ×4 (data-active / multiplexer window not found), palette-polish touch ×2 (label width 53.99 vs 54), toast-stack ×2 (behind.width equals mid.width). Log: ~/j383-c/base-specs.log. Not investigated further.

Findings (visible side effects of the owner's answers):

  1. On a phone the details popover's key/value rows now wrap ("8 logical cores · AMD EPYC 7813" breaks onto two lines) and the popover grows by about 20 px, because the phone body is 0.9375 rem (15 px) instead of 0.875 rem. See the 393×852 sheet, popover row, last two columns.
  2. The phone rename field stays 16 px (iOS) while the label it replaces is now 13 px: a 3 px step while renaming. Needs an owner call if it matters.
  3. The 1.75 rem cluster and mesh frames were not rounded (no step above --r-lg).

Contact sheets (phones first; columns: part 1 ink, part 2 ink, part 1 paper, part 2 paper; rows: canvas, sidebar, palette, popover, sign-in; Settings is not captured on phones): p2-393x852-before-after.png, p2-375x667-before-after.png, p2-1440x900-before-after.png. Overlay pairs: pairs-393x852-before-after.png, pairs-375x667-before-after.png, pairs-1440x900-before-after.png (same column order, rows are pairs).

Open questions (unchanged from the progress comment): phone sizes read as unscaled rem (body 15, labels 13); title 1.25 rem (22 px at the 110 % root, 20 px on a phone); --text-* kept as aliases.

**J383 part 2 finish: `ui-consistency-d372`, head `af1dc7db324ba2c9b5cd29cc508c0c5ec30f6521`** (13 commits stacked on part 1 head `c27a7ad2d`). Nothing merged, nothing deployed. Commits: durations (3 steps), `EASE_OUT`, `--r-xs` and radius rounding, one control height, `--ui-icon-sm`, grid (3/5 px round to 4), type roles, one title role, docs.css, six-step layer scale, overlay-pair spec, TASTE, doc comments (D345). **`scripts/verify quick` on the head, verbatim:** ``` +---------------------------+--------------+ | Rust gate | Result | +---------------------------+--------------+ | scope guard | PASS | | migration names | PASS | | Nix Cargo vendoring | PASS | | fmt | PASS | | doc coverage ratchet | PASS | | shape lints off (D351) | PASS | | doc comments on diff | PASS | | fixed test waits | PASS | | ui literals | PASS | | tracer typecheck | PASS | +---------------------------+--------------+ QUICK pre-check passed. This is not the gate: run scripts/verify fast before reporting done. QUICK_EXIT=0 ``` **Bundle budget** (`bun run build.ts && bun run transfer-budget.ts`, exit 0, no cap touched): critical gzip 319,018 / 327,680 (part 1: 319,007), deferred gzip 58,560 / 61,440 (58,546), deferred brotli 51,916 / 55,296 (51,933). Log: `~/j383-p2b/j383-p2-budget.log` on dev-02. **Census (production build, 1440×900, 393×852, 375×667, ink and paper, 6 screens), part 1 head vs part 2:** - transition durations: 9 values (0.12 to 0.26 s) become 3 (0.12 s ×1398, 0.18 s ×480, 0.24 s ×56); - z-index: 5→20, 6→30, 4→10, 7→31, 8→32, 12→35, and 0, 1, 2, 3, −1 unchanged (monotone, as designed); - radii and font sizes: the old touch body (16 px), secondary (15 px) and caption (13 px) are gone; the off-scale radii (5.5, 9, 10, 15, 16, 24.2 px etc.) are gone. **Overlay pairs** (`ui-layer-pairs.spec.ts`, hit test with `elementFromPoint`): before and after both 28 of 30 pairs have the upper layer on top, with the same result in every cell. The 2 that fail in both runs are `tooltip-over-sidebar` at 1440×900 (ink and paper): my driver could not make the tooltip appear, so that pair is **not** proven by capture (the tooltip layers map 1000/1001/1002 to `--z-tooltip` + 0/1/2). Phones skip Settings and hover-tooltip pairs (Settings is not reachable on a phone in the mock). **Browser specs on a production build (dev-02):** docs, docs-reading, docs-palette, icon-text-centre, terminal-links, toast-stack, tooltip-fit, settings-rows, settings-edges, sidebar-polish, palette-polish, palette-navigation, row-menu, addhost-wizard, replacement-alignment, lock-icons: 165 passed in the first run. 13 failed; the rerun after my fixes: - **Fixed (specs that pinned old px values; the new values follow D372):** docs-reading and docs (desktop: 17 px icon → 18.7, 1.5rem and 16 px offsets are now root relative), replacement-alignment (a touch rename field stays 16 px for iOS, above the 13 px label, so the spec allows `max(label, 16)` on touch). Reruns: docs and docs-reading 9 passed; replacement-alignment and icon-text-centre green in the combined rerun (log `~/j383-c/p2-specs.log`). - **Pre-existing, identical on part 1 head (and so not caused by D372):** docs-palette ×4 (`data-active` / multiplexer window not found), palette-polish touch ×2 (label width 53.99 vs 54), toast-stack ×2 (`behind.width` equals `mid.width`). Log: `~/j383-c/base-specs.log`. Not investigated further. **Findings (visible side effects of the owner's answers):** 1. On a phone the details popover's key/value rows now wrap ("8 logical cores · AMD EPYC 7813" breaks onto two lines) and the popover grows by about 20 px, because the phone body is 0.9375 rem (15 px) instead of 0.875 rem. See the 393×852 sheet, popover row, last two columns. 2. The phone rename field stays 16 px (iOS) while the label it replaces is now 13 px: a 3 px step while renaming. Needs an owner call if it matters. 3. The 1.75 rem cluster and mesh frames were not rounded (no step above `--r-lg`). **Contact sheets** (phones first; columns: part 1 ink, part 2 ink, part 1 paper, part 2 paper; rows: canvas, sidebar, palette, popover, sign-in; Settings is not captured on phones): `p2-393x852-before-after.png`, `p2-375x667-before-after.png`, `p2-1440x900-before-after.png`. Overlay pairs: `pairs-393x852-before-after.png`, `pairs-375x667-before-after.png`, `pairs-1440x900-before-after.png` (same column order, rows are pairs). Open questions (unchanged from the progress comment): phone sizes read as unscaled rem (body 15, labels 13); title 1.25 rem (22 px at the 110 % root, 20 px on a phone); `--text-*` kept as aliases.
p2-393x852-before-after.png
5 MiB
p2-375x667-before-after.png
4.4 MiB
p2-1440x900-before-after.png
2.9 MiB
pairs-393x852-before-after.png
7.7 MiB
pairs-375x667-before-after.png
6.6 MiB
pairs-1440x900-before-after.png
3.7 MiB
p2-393x852-before-after.png p2-375x667-before-after.png p2-1440x900-before-after.png pairs-393x852-before-after.png pairs-375x667-before-after.png pairs-1440x900-before-after.png
kayg commented 2026-11-09 00:34:47 +00:00
Author
Owner
Copy link

J401 start (independent UI review of J383), 2026-11-09.

Targets: part 1 ui-consistency-j195-deferred-rebased @ c27a7ad2de8d196b28d85befca777d4bf3d11dbf; part 2 ui-consistency-d372 @ af1dc7db324ba2c9b5cd29cc508c0c5ec30f6521 (stacked, 13 commits). origin/dev is 44e80ccc5.

Note for the record: D372 is not in DECISIONS.md on any ref (origin/dev, origin/main, the part 2 branch: git grep D372 -- DECISIONS.md finds nothing). TASTE on the part 2 branch cites it. I review against the owner's answers in comment 86933 on this issue, which is the only record, so the decision needs writing into DECISIONS.md by the owner.

Plan: (1) part 1 before-vs-before noise control, then before-vs-after, 3 sizes x ink/paper; (2) D372 item by item with computed values; (3) the 8 failing specs on unmodified origin/dev; (4) tooltip-over-sidebar proof, popover growth, rename field; (5) transfer-budget, verify quick, and verify fast on compute01 for the stacked head merged with origin/dev. Browser work on dev-02 through build-slot. Estimate about 4 h.

**J401 start (independent UI review of J383), 2026-11-09.** Targets: part 1 `ui-consistency-j195-deferred-rebased` @ `c27a7ad2de8d196b28d85befca777d4bf3d11dbf`; part 2 `ui-consistency-d372` @ `af1dc7db324ba2c9b5cd29cc508c0c5ec30f6521` (stacked, 13 commits). origin/dev is `44e80ccc5`. Note for the record: **D372 is not in `DECISIONS.md` on any ref** (origin/dev, origin/main, the part 2 branch: `git grep D372 -- DECISIONS.md` finds nothing). TASTE on the part 2 branch cites it. I review against the owner's answers in comment 86933 on this issue, which is the only record, so the decision needs writing into DECISIONS.md by the owner. Plan: (1) part 1 before-vs-before noise control, then before-vs-after, 3 sizes x ink/paper; (2) D372 item by item with computed values; (3) the 8 failing specs on unmodified origin/dev; (4) tooltip-over-sidebar proof, popover growth, rename field; (5) `transfer-budget`, `verify quick`, and `verify fast` on compute01 for the stacked head merged with origin/dev. Browser work on dev-02 through build-slot. Estimate about 4 h.
kayg commented 2026-11-09 00:44:43 +00:00
Author
Owner
Copy link

J401 finding 1 (static review of part 2 af1dc7db3; evidence so far is source, runtime numbers follow).

  1. Phone sizes in rem are right. ui-scale.ts and styles.css set the root to 16px x --ui-base x --ui-scale, and --ui-base is 1 on touch. So on a phone 1 rem is 16 px at the default slider: --ui-size-body: 0.9375rem renders 15 px and --ui-size-label: 0.8125rem renders 13 px, exactly the owner's numbers. Writing them in px would break the UI size slider (TASTE section 2, "every length is in rem"). Verdict on call 1: correct. On desktop the same tokens render 15.4 / 14.3 px, which is the existing sign-in card size, and the owner's answer names phones only.
  2. z-index map is strictly monotone (checked by script over the diff of 86d46ec2f). old 1,2,3 -> 1,2,3; 4 -> 10; 5 -> 20; 6..10 -> 30..34; 12 -> 35; 20 -> 36; 30 -> 37; 90 -> 38; 900,901 -> 60,61; 1000..1002 -> 70..72. No two old values collapse and none change order.
  3. Leftover non-token durations (D372 says three durations). transition declarations in CSS moved, but these still write their own number: row-menu.ts:111 150 ms, branched.ts:258/281 150 and 280 ms, capsule-tabs.ts:71 200 ms, main.ts:409 240 ms (equals --t-panel but a literal), window-resize.ts 220 ms, lock-toggle.ts:114 220 ms, auth-motion.ts:41 260 ms, style.css:672 keytray 280 ms, addhost.css 340 and 420 ms, .ui-sidebar-card[data-morph] 60 ms. J383 said animations and keyframes were left on purpose, so this is a scope call, not a bug; it should be listed in the TASTE row so the "A new rule never writes a number" sentence is not read as already true.
  4. TASTE row 46 now contradicts the phone result. It says "Phone values are 15-16 px or larger, keys 13-14 px or larger". After D372 the phone has eyebrow/secondary 12 px and body-in-row 13 px (was 16 / 15 / 13 px). The owner's numbers (labels 13 px) are the later decision, so the code is right and the TASTE row needs the owner to amend it; I do not change decisions.
  5. DECISIONS.md has no D372 entry on any ref (also noted in my start comment).
**J401 finding 1 (static review of part 2 `af1dc7db3`; evidence so far is source, runtime numbers follow).** 1. **Phone sizes in rem are right.** `ui-scale.ts` and `styles.css` set the root to `16px x --ui-base x --ui-scale`, and `--ui-base` is 1 on touch. So on a phone 1 rem is 16 px at the default slider: `--ui-size-body: 0.9375rem` renders 15 px and `--ui-size-label: 0.8125rem` renders 13 px, exactly the owner's numbers. Writing them in px would break the UI size slider (TASTE section 2, "every length is in rem"). Verdict on call 1: **correct**. On desktop the same tokens render 15.4 / 14.3 px, which is the existing sign-in card size, and the owner's answer names phones only. 2. **z-index map is strictly monotone (checked by script over the diff of `86d46ec2f`).** old 1,2,3 -> 1,2,3; 4 -> 10; 5 -> 20; 6..10 -> 30..34; 12 -> 35; 20 -> 36; 30 -> 37; 90 -> 38; 900,901 -> 60,61; 1000..1002 -> 70..72. No two old values collapse and none change order. 3. **Leftover non-token durations (D372 says three durations).** `transition` declarations in CSS moved, but these still write their own number: `row-menu.ts:111` 150 ms, `branched.ts:258/281` 150 and 280 ms, `capsule-tabs.ts:71` 200 ms, `main.ts:409` 240 ms (equals `--t-panel` but a literal), `window-resize.ts` 220 ms, `lock-toggle.ts:114` 220 ms, `auth-motion.ts:41` 260 ms, `style.css:672` keytray 280 ms, `addhost.css` 340 and 420 ms, `.ui-sidebar-card[data-morph]` 60 ms. J383 said animations and keyframes were left on purpose, so this is a scope call, not a bug; it should be listed in the TASTE row so the "A new rule never writes a number" sentence is not read as already true. 4. **TASTE row 46 now contradicts the phone result.** It says "Phone values are 15-16 px or larger, keys 13-14 px or larger". After D372 the phone has eyebrow/secondary 12 px and body-in-row 13 px (was 16 / 15 / 13 px). The owner's numbers (labels 13 px) are the later decision, so the code is right and the TASTE row needs the owner to amend it; I do not change decisions. 5. **`DECISIONS.md` has no D372 entry** on any ref (also noted in my start comment).
kayg commented 2026-11-09 00:57:32 +00:00
Author
Owner
Copy link

J401 finding 2: part 1 pixel identity measured with a noise control (production builds, dev-02).

Method: the part 1 base is a894dd877. Two captures of that same commit (baseA, baseB) are the noise control; part 1 head c27a7ad2d is the third run. 32 screenshots per run (5 screens on phones, 6 on desktop; 1440x900, 393x852, 375x667, ink and paper). A pixel counts as changed when a channel moves by more than 8/255 (pdiff.mjs, log ~/j401/diff-control.txt and diff-p1.txt on dev-02).

pair images with any pixel over 8/255 worst image largest changed-pixel count
baseA vs baseB (control, same commit) 14 of 32 0.024 % 557 px
baseA vs part 1 10 of 32 0.024 % 557 px

Every changed region in both pairs is the same small box: the clock text of the mock terminal (bottom right of the canvas) and a few antialiased pixels in popovers. The sign-in images differ by 3k to 90k pixels at a delta of 1 to 2 out of 255 in both pairs (the card's backdrop fade). Part 1 is not worse than two runs of unchanged code.

Computed census (cdiff.mjs): 846 cells (font size, weight, line height, letter spacing, radius, transition duration and easing, z-index, per screen and size) compared; baseA vs baseB 0 differ, baseA vs part 1 0 differ. The only change is the probes, in all 6 census files: .check goes 0s | ease -> 0.14s | cubic-bezier(0.23, 1, 0.32, 1) and .swap svg 0s | ease -> 0.15s | ... on canvas, sidebar, palette and popover. That is the --ease fix and nothing else.

Part 1: pixel-identical except --ease, proven. No change needed.

8-spec attribution, first half (unmodified origin/dev 44e80ccc5, same spec files, production build): 34 passed, 6 failed. docs-palette x4 and toast-stack x2 fail on unmodified dev too, so they are not this stack's fault. palette-polish touch x2 pass on dev; I am running base a894dd877 and part 1 to find where they break. Log ~/j401/logs/devspecs.log.

**J401 finding 2: part 1 pixel identity measured with a noise control (production builds, dev-02).** Method: the part 1 base is `a894dd877`. Two captures of that same commit (baseA, baseB) are the noise control; part 1 head `c27a7ad2d` is the third run. 32 screenshots per run (5 screens on phones, 6 on desktop; 1440x900, 393x852, 375x667, ink and paper). A pixel counts as changed when a channel moves by more than 8/255 (`pdiff.mjs`, log `~/j401/diff-control.txt` and `diff-p1.txt` on dev-02). | pair | images with any pixel over 8/255 | worst image | largest changed-pixel count | |---|---|---|---| | baseA vs baseB (control, same commit) | 14 of 32 | 0.024 % | 557 px | | baseA vs part 1 | 10 of 32 | 0.024 % | 557 px | Every changed region in both pairs is the same small box: the clock text of the mock terminal (bottom right of the canvas) and a few antialiased pixels in popovers. The sign-in images differ by 3k to 90k pixels at a delta of 1 to 2 out of 255 in both pairs (the card's backdrop fade). Part 1 is not worse than two runs of unchanged code. Computed census (`cdiff.mjs`): 846 cells (font size, weight, line height, letter spacing, radius, transition duration and easing, z-index, per screen and size) compared; baseA vs baseB 0 differ, baseA vs part 1 **0 differ**. The only change is the probes, in all 6 census files: `.check` goes `0s | ease` -> `0.14s | cubic-bezier(0.23, 1, 0.32, 1)` and `.swap svg` `0s | ease` -> `0.15s | ...` on canvas, sidebar, palette and popover. That is the `--ease` fix and nothing else. **Part 1: pixel-identical except `--ease`, proven. No change needed.** **8-spec attribution, first half (unmodified origin/dev `44e80ccc5`, same spec files, production build):** 34 passed, 6 failed. `docs-palette` x4 and `toast-stack` x2 **fail on unmodified dev too**, so they are not this stack's fault. `palette-polish` touch x2 **pass on dev**; I am running base `a894dd877` and part 1 to find where they break. Log `~/j401/logs/devspecs.log`.
kayg commented 2026-11-09 01:31:31 +00:00
Author
Owner
Copy link

J401 finding 3: tooltip-over-sidebar is now proven, and what the part 2 census says (production builds, dev-02).

Tooltip over the open sidebar (1440x900, ink and paper). J383's driver hovered a button whose tooltip never showed, and the product's .tip-pos is pointer-events: none, so elementFromPoint can never return a tooltip even when it is on top. My probe (j401-tooltip.spec.ts, not committed) hovers every sidebar control with a real mouse move, waits for the visible tooltip, requires that its box overlaps #sidebar, makes the tooltip hittable for the probe only, and hit-tests the centre of its surface.

build tooltips shown overlap with sidebar tooltip z / sidebar z hit inside tooltip
part 1 c27a7ad2d "Finished successfully", "Detached", "New output", "Detached" 4853, 2250, 3227, 2250 px² 1002 / 5 4 of 4
part 2 af1dc7db3 the same four the same four 72 / 20 4 of 4

Ink and paper both pass. Screenshot: the "Finished successfully" tooltip straddles the sidebar edge and is drawn over the rows (~/j401/ev/tip-p2/j401-tip/tip-p2/ink-3.png). 4 of the 8 sidebar controls show no tooltip (they have no label to show); that is the same on both builds. Pair proven.

Part 2 census against part 1 (393x852 and 1440x900 ink, all screens merged; ~/j401/vals.mjs):

  • Phone font sizes: 16 px x181 and 15 px x89 and 14 px x17 are gone; 13 px 956 -> 965, 12 px 5 -> 276, 15 px 89 -> 18 (body role). Sign-in title 23 -> 20 px.
  • Desktop: 16.5 and 23.1 px gone, the title is 22 px (x1), 12.1 px x269 -> 13.2 px x384.
  • Radii: 5, 16, 20 px (phone) and 4.4, 5.5, 11, 15.4, 17.6, 24.2 px (desktop) are gone; no new value appears. 28 / 30.8 px (the 1.75 rem frames) remain.
  • Durations: 9 values -> exactly 3 (0.12 / 0.18 / 0.24 s) in both sizes.
  • z-index: 5 -> 20, 6 -> 30, 4 -> 10 etc., same element counts per layer.
  • Weights unchanged (450/500/550/600/650).

What shrank on a phone, with elements (this is the cost of mapping --text-body to the label size): 148 palette result labels, 31 sidebar row names and 2 capsule names go 16 -> 13 px; 87 palette result icons' text and the group labels and hints go 15 -> 12 px. These are the primary tap targets on an iPhone. See the verdict on this in the final report.

**J401 finding 3: `tooltip-over-sidebar` is now proven, and what the part 2 census says (production builds, dev-02).** **Tooltip over the open sidebar (1440x900, ink and paper).** J383's driver hovered a button whose tooltip never showed, and the product's `.tip-pos` is `pointer-events: none`, so `elementFromPoint` can never return a tooltip even when it is on top. My probe (`j401-tooltip.spec.ts`, not committed) hovers every sidebar control with a real mouse move, waits for the visible tooltip, requires that its box overlaps `#sidebar`, makes the tooltip hittable for the probe only, and hit-tests the centre of its surface. | build | tooltips shown | overlap with sidebar | tooltip z / sidebar z | hit inside tooltip | |---|---|---|---|---| | part 1 `c27a7ad2d` | "Finished successfully", "Detached", "New output", "Detached" | 4853, 2250, 3227, 2250 px² | 1002 / 5 | 4 of 4 | | part 2 `af1dc7db3` | the same four | the same four | 72 / 20 | 4 of 4 | Ink and paper both pass. Screenshot: the "Finished successfully" tooltip straddles the sidebar edge and is drawn over the rows (`~/j401/ev/tip-p2/j401-tip/tip-p2/ink-3.png`). 4 of the 8 sidebar controls show no tooltip (they have no label to show); that is the same on both builds. **Pair proven.** **Part 2 census against part 1 (393x852 and 1440x900 ink, all screens merged; `~/j401/vals.mjs`):** - Phone font sizes: 16 px x181 and 15 px x89 and 14 px x17 are gone; 13 px 956 -> 965, 12 px 5 -> 276, 15 px 89 -> 18 (body role). Sign-in title 23 -> 20 px. - Desktop: 16.5 and 23.1 px gone, the title is 22 px (x1), 12.1 px x269 -> 13.2 px x384. - Radii: 5, 16, 20 px (phone) and 4.4, 5.5, 11, 15.4, 17.6, 24.2 px (desktop) are gone; no new value appears. 28 / 30.8 px (the 1.75 rem frames) remain. - Durations: 9 values -> exactly 3 (0.12 / 0.18 / 0.24 s) in both sizes. - z-index: 5 -> 20, 6 -> 30, 4 -> 10 etc., same element counts per layer. - Weights unchanged (450/500/550/600/650). **What shrank on a phone, with elements (this is the cost of mapping `--text-body` to the label size):** 148 palette result labels, 31 sidebar row names and 2 capsule names go 16 -> 13 px; 87 palette result icons' text and the group labels and hints go 15 -> 12 px. These are the primary tap targets on an iPhone. See the verdict on this in the final report.
kayg commented 2026-11-09 02:33:54 +00:00
Author
Owner
Copy link

J401 finding 4: the 8 failing specs, attribution (all runs on production builds, dev-02; logs ~/j401/logs/devspecs.log, pp.log, pp2.log, aw.log).

spec unmodified origin/dev 44e80ccc5 base a894dd877 part 1 verdict
docs-palette x4 fails 4 of 4 (data-active never set on the multiplexer window) n/a fails not this stack's fault
toast-stack x2 fails 2 of 2 (behind.width equals mid.width on phone, mid.y over f.y by 0.38 px on desktop) n/a fails not this stack's fault
palette-polish touch x2 passes when run with the other two files, fails 2 of 2 when run alone (-g touch), both 393 and 375, twice (label "deploy" width 57.866 against 58) same: fails twice 375 fails not this stack's fault: it is order-dependent on dev itself, the value 57.866 is identical on every build

So no commit of this stack broke any of the 8; there was nothing to bisect. palette-polish touch should get its own issue (it passes or fails with what ran before it).

One more thing I hit and ruled out: running 13 spec files together on part 2 failed all 42 addhost-wizard cases with .cl-mockup not found. Run alone, addhost-wizard on part 2 passes 58 of 58 (11.8 min). That was shared test-double state between spec files in one run, not the stack. In the same combined run 100 passed, 8 skipped and only docs-palette x4 (the dev failure above) failed besides the addhost cascade.

Side effects, measured on the phone (393x852 and 375x667 touch contexts, DPR 3, popt-p1/popt-p2):

  • Details popover: the box is the same height on both builds (652 px and 467 px; it is capped by the screen and scrolls). What grows is the content: the CPU row "8 logical cores · AMD EPYC 7B13" wraps on part 2 at 393 (0 wrapped rows before, 1 now) and "Embedded · user" also wraps at 375 (1 before, 2 now). Each wrap is about 20 px more scroll. The wrap leaves the orphan "7B13" alone on a second line (screenshot ~/j401/ev/popt-p2/j401-pop/popt-p2/popover-393x852.png). That is a TASTE annoyance (a value breaking into a one-word line), not a bug of the token change.
  • Rename field: stays 16 px while the sidebar row label is now 13 px, a 3 px jump while renaming. TASTE says a replacement starts at the label's x, on its centre line and in its size. J383 loosened replacement-alignment.spec.ts to touch ? max(label, 16) : label.
**J401 finding 4: the 8 failing specs, attribution (all runs on production builds, dev-02; logs `~/j401/logs/devspecs.log`, `pp.log`, `pp2.log`, `aw.log`).** | spec | unmodified origin/dev `44e80ccc5` | base `a894dd877` | part 1 | verdict | |---|---|---|---|---| | docs-palette x4 | fails 4 of 4 (`data-active` never set on the multiplexer window) | n/a | fails | **not this stack's fault** | | toast-stack x2 | fails 2 of 2 (`behind.width` equals `mid.width` on phone, `mid.y` over `f.y` by 0.38 px on desktop) | n/a | fails | **not this stack's fault** | | palette-polish touch x2 | passes when run with the other two files, **fails 2 of 2 when run alone (`-g touch`), both 393 and 375, twice** (label "deploy" width 57.866 against 58) | same: fails twice | 375 fails | **not this stack's fault**: it is order-dependent on dev itself, the value 57.866 is identical on every build | So no commit of this stack broke any of the 8; there was nothing to bisect. `palette-polish` touch should get its own issue (it passes or fails with what ran before it). One more thing I hit and ruled out: running 13 spec files together on part 2 failed all 42 `addhost-wizard` cases with `.cl-mockup` not found. Run alone, `addhost-wizard` on part 2 passes 58 of 58 (11.8 min). That was shared test-double state between spec files in one run, not the stack. In the same combined run 100 passed, 8 skipped and only `docs-palette` x4 (the dev failure above) failed besides the addhost cascade. **Side effects, measured on the phone (393x852 and 375x667 touch contexts, DPR 3, `popt-p1`/`popt-p2`):** - Details popover: the box is the same height on both builds (652 px and 467 px; it is capped by the screen and scrolls). What grows is the content: the CPU row "8 logical cores · AMD EPYC 7B13" wraps on part 2 at 393 (0 wrapped rows before, 1 now) and "Embedded · user" also wraps at 375 (1 before, 2 now). Each wrap is about 20 px more scroll. The wrap leaves the orphan "7B13" alone on a second line (screenshot `~/j401/ev/popt-p2/j401-pop/popt-p2/popover-393x852.png`). That is a TASTE annoyance (a value breaking into a one-word line), not a bug of the token change. - Rename field: stays 16 px while the sidebar row label is now 13 px, a 3 px jump while renaming. TASTE says a replacement starts at the label's x, on its centre line and in its size. J383 loosened `replacement-alignment.spec.ts` to `touch ? max(label, 16) : label`.
kayg commented 2026-11-09 02:58:53 +00:00
Author
Owner
Copy link

J401 final report. Part 1: APPROVE. Part 2: REQUEST CHANGES (one small code change, two doc/spec changes). Nothing merged, nothing deployed.

SHAs: part 1 c27a7ad2de8d196b28d85befca777d4bf3d11dbf, part 2 af1dc7db324ba2c9b5cd29cc508c0c5ec30f6521, origin/dev 44e80ccc5 (merge with part 2 is conflict-free: b54a1100b, local only, gate worktree on compute01).

Part 1 (APPROVE). Pixel identity proven against a same-commit noise control (worst image 0.024 % in both; 846 census cells, 0 differ; only the --ease probes change). Details in finding 2.

Part 2, D372 item by item (owner's answers: comment 86933; D372 is not in DECISIONS.md, the owner must add it):

  1. Phone body 15 / labels 13, set once: done. Touch block sets --ui-size-body: 0.9375rem; labels 0.8125 rem everywhere. rem is correct (phone root is 16 px, so 15 and 13 px render exactly; px would break the UI-size slider).
  2. --r-xs with nested radii: done; 5, 16, 20 px and 4.4/5.5/11/15.4/17.6/24.2 px radii are gone from the census. The 1.75 rem frames (.cl-region, .cl-ghost, .cl-chip: 28 / 30.8 px) left alone: right (outer frames around 1.125 rem windows, not off-scale one-offs), but they need a one-line "nested outer frame" comment and an audit-doc row, and have neither.
  3. 2 px half steps: done (one calc(3px - 0.125rem) remains in styles.css:1135).
  4. One control height: done (--ui-row-h gone; 0 reads).
  5. --ui-icon-sm: partly. Still literal glyph sizes: .bm-icon svg 15 px (styles.css:757), .bm-icon 18 px (style.css:468, fights the token one), .ah-hero .ah-icon 18 px, .ui-source-icon 13 px in the add-host list. TASTE now says "always one of the two tokens".
  6. Three durations: CSS done (census 9 values -> 3); TypeScript animations and several CSS ones are left (list in finding 1). Fine as scope, but TASTE's "a transition lasts ..." should say CSS transitions.
  7. Docs on the app root size: done, phone body 15 px, desktop 15.4 px.
  8. 22 px title: desktop 22 px; phone 20 px. Wrong on the phone. Old phone title was 23 px, so 20 px is smaller than any title before, and the owner said 22. Fix (one line): --ui-size-title: 1.375rem in the touch block next to --ui-size-body (22 px on the 16 px root; desktop stays 1.25 rem = 22 px). Same pattern the owner already approved for body.
  9. z-index scale: done, map strictly monotone; tooltip-over-sidebar proven (finding 3).
  10. EASE_OUT: done (packages/ui/src/motion.ts, 14 call sites, export added).

J383's calls: rem vs px: right. Title 1.25 rem: right for desktop, wrong for phones (above). 1.75 rem frames: right, add the comment.

Visible side effects: (a) Phone row text shrinks: 148 palette result labels and 31 sidebar names go 16 -> 13 px, secondary text 15 -> 12 px, because --text-body now aliases the label size. It follows the owner's "labels 13 px", so I do not block it, but it contradicts TASTE row 46 ("phone 15-16 px or larger"); the owner should see the phone sheet and amend row 46. (b) Popover: box height is unchanged (652 / 467 px, capped, scrolls); content grows about 20 px per wrapped row, and the CPU row leaves the orphan "7B13" on its own line. Not blocking; a text-wrap: balance-style fix belongs in a follow-up. (c) Rename field: 16 px against a 13 px label is required by iOS (no zoom) and is correct, but J383 loosened replacement-alignment.spec.ts silently. Required: assert exactly 16 px on touch and record the exception in TASTE's replacement row.

8-spec attribution: none of them is this stack's fault. docs-palette x4 and toast-stack x2 fail on unmodified origin/dev; palette-polish touch x2 fail on origin/dev and on base a894dd877 when run alone (order-dependent, identical 57.866 value). Details in finding 4.

Gates:

  • bun run build.ts && bun run transfer-budget.ts on part 2: exit 0, no cap edited (transfer-budget.ts unchanged): critical gzip 319008/327680, brotli 266411/273408; deferred gzip 58553/61440, brotli 52010/55296. Log ~/j401/logs/gates.log on dev-02.
  • scripts/verify quick on a clean checkout of part 2 (dev-02, base origin/dev), verbatim: scope guard, migration names, Nix Cargo vendoring, fmt, doc coverage ratchet, shape lints off (D351), doc comments on diff, fixed test waits, ui literals, tracer typecheck: all PASS, QUICK_EXIT=0 (~/j401/logs/quick.log). (A first run failed only on doc comments for my own uncommitted probe specs; removed.)
  • scripts/verify fast on compute01 on part 2 merged with origin/dev 44e80ccc5 (b54a1100b): every Rust-gate row PASS (Nix Cargo vendoring, fmt, migration names, migration tooling tests, VT native fmt, CLI plugin compile-out, clippy, SSH runtime test helper, real-binary fixture paths, Traefik component manifest, release component publication, test, SSH inode reuse fixture, shim process-per-user, cargo-deny, gate architecture, doc coverage ratchet, shape lints off, doc comments on diff, fixed test waits, probe release exclusion, nightly fuzz reporter, static builds 1856 s), privileged integration SKIP (not root), then the tracer budget and TypeScript tests ran with 50 pass, 0 fail. My wrapper did not record the process exit code, so the exit status is inferred from the absence of any FAIL row. Log ~/j401/verify-fast.log on compute01.

Required before part 2 merges: (1) phone title 1.375 rem; (2) tighten the rename assertion to 16 px and record the iOS exception in TASTE; (3) comment the 1.75 rem frames and make the icon and duration residue explicit (fix the four icon sizes, or list them as exceptions). For the owner, not me: add D372 to DECISIONS.md; amend TASTE row 46.

**J401 final report. Part 1: APPROVE. Part 2: REQUEST CHANGES (one small code change, two doc/spec changes).** Nothing merged, nothing deployed. SHAs: part 1 `c27a7ad2de8d196b28d85befca777d4bf3d11dbf`, part 2 `af1dc7db324ba2c9b5cd29cc508c0c5ec30f6521`, origin/dev `44e80ccc5` (merge with part 2 is conflict-free: `b54a1100b`, local only, gate worktree on compute01). **Part 1 (APPROVE).** Pixel identity proven against a same-commit noise control (worst image 0.024 % in both; 846 census cells, 0 differ; only the `--ease` probes change). Details in finding 2. **Part 2, D372 item by item (owner's answers: comment 86933; D372 is not in DECISIONS.md, the owner must add it):** 1. Phone body 15 / labels 13, set once: **done**. Touch block sets `--ui-size-body: 0.9375rem`; labels 0.8125 rem everywhere. rem is correct (phone root is 16 px, so 15 and 13 px render exactly; px would break the UI-size slider). 2. `--r-xs` with nested radii: **done**; 5, 16, 20 px and 4.4/5.5/11/15.4/17.6/24.2 px radii are gone from the census. The 1.75 rem frames (`.cl-region`, `.cl-ghost`, `.cl-chip`: 28 / 30.8 px) left alone: **right** (outer frames around 1.125 rem windows, not off-scale one-offs), but they need a one-line "nested outer frame" comment and an audit-doc row, and have neither. 3. 2 px half steps: **done** (one `calc(3px - 0.125rem)` remains in `styles.css:1135`). 4. One control height: **done** (`--ui-row-h` gone; 0 reads). 5. `--ui-icon-sm`: **partly**. Still literal glyph sizes: `.bm-icon svg` 15 px (`styles.css:757`), `.bm-icon` 18 px (`style.css:468`, fights the token one), `.ah-hero .ah-icon` 18 px, `.ui-source-icon` 13 px in the add-host list. TASTE now says "always one of the two tokens". 6. Three durations: **CSS done** (census 9 values -> 3); TypeScript animations and several CSS ones are left (list in finding 1). Fine as scope, but TASTE's "a transition lasts ..." should say CSS transitions. 7. Docs on the app root size: **done**, phone body 15 px, desktop 15.4 px. 8. 22 px title: **desktop 22 px; phone 20 px. Wrong on the phone.** Old phone title was 23 px, so 20 px is smaller than any title before, and the owner said 22. Fix (one line): `--ui-size-title: 1.375rem` in the touch block next to `--ui-size-body` (22 px on the 16 px root; desktop stays 1.25 rem = 22 px). Same pattern the owner already approved for body. 9. z-index scale: **done**, map strictly monotone; `tooltip-over-sidebar` proven (finding 3). 10. `EASE_OUT`: **done** (`packages/ui/src/motion.ts`, 14 call sites, export added). **J383's calls:** rem vs px: right. Title 1.25 rem: right for desktop, wrong for phones (above). 1.75 rem frames: right, add the comment. **Visible side effects:** (a) Phone row text shrinks: 148 palette result labels and 31 sidebar names go 16 -> 13 px, secondary text 15 -> 12 px, because `--text-body` now aliases the label size. It follows the owner's "labels 13 px", so I do not block it, but it contradicts TASTE row 46 ("phone 15-16 px or larger"); the owner should see the phone sheet and amend row 46. (b) Popover: box height is unchanged (652 / 467 px, capped, scrolls); content grows about 20 px per wrapped row, and the CPU row leaves the orphan "7B13" on its own line. Not blocking; a `text-wrap: balance`-style fix belongs in a follow-up. (c) Rename field: 16 px against a 13 px label is required by iOS (no zoom) and is correct, but J383 loosened `replacement-alignment.spec.ts` silently. Required: assert exactly 16 px on touch and record the exception in TASTE's replacement row. **8-spec attribution:** none of them is this stack's fault. docs-palette x4 and toast-stack x2 fail on unmodified origin/dev; palette-polish touch x2 fail on origin/dev and on base `a894dd877` when run alone (order-dependent, identical 57.866 value). Details in finding 4. **Gates:** - `bun run build.ts && bun run transfer-budget.ts` on part 2: exit 0, no cap edited (`transfer-budget.ts` unchanged): critical gzip 319008/327680, brotli 266411/273408; deferred gzip 58553/61440, brotli 52010/55296. Log `~/j401/logs/gates.log` on dev-02. - `scripts/verify quick` on a clean checkout of part 2 (dev-02, base origin/dev), verbatim: scope guard, migration names, Nix Cargo vendoring, fmt, doc coverage ratchet, shape lints off (D351), doc comments on diff, fixed test waits, ui literals, tracer typecheck: all PASS, `QUICK_EXIT=0` (`~/j401/logs/quick.log`). (A first run failed only on doc comments for my own uncommitted probe specs; removed.) - `scripts/verify fast` on compute01 on part 2 merged with origin/dev `44e80ccc5` (`b54a1100b`): every Rust-gate row PASS (Nix Cargo vendoring, fmt, migration names, migration tooling tests, VT native fmt, CLI plugin compile-out, clippy, SSH runtime test helper, real-binary fixture paths, Traefik component manifest, release component publication, test, SSH inode reuse fixture, shim process-per-user, cargo-deny, gate architecture, doc coverage ratchet, shape lints off, doc comments on diff, fixed test waits, probe release exclusion, nightly fuzz reporter, static builds 1856 s), `privileged integration` SKIP (not root), then the tracer budget and TypeScript tests ran with 50 pass, 0 fail. My wrapper did not record the process exit code, so the exit status is inferred from the absence of any FAIL row. Log `~/j401/verify-fast.log` on compute01. **Required before part 2 merges:** (1) phone title 1.375 rem; (2) tighten the rename assertion to 16 px and record the iOS exception in TASTE; (3) comment the 1.75 rem frames and make the icon and duration residue explicit (fix the four icon sizes, or list them as exceptions). **For the owner, not me:** add D372 to DECISIONS.md; amend TASTE row 46.
kayg commented 2026-11-09 04:34:06 +00:00
Author
Owner
Copy link

J383 part 2, review round 1 done (J401's list): ui-consistency-d372 rebased on dev 966c52e07, head 24b7c21599ec262a647287ab3260314e2ff3ed88 (20 commits; part 1 was already in dev). Not merged.

  1. Phone title: --ui-size-title: 1.375rem in the touch block next to --ui-size-body; desktop stays 1.25 rem.
  2. Icons: .bm-icon (18 px box) and .ah-hero .ah-icon read --ui-icon; .bm-icon svg (15 px) and the add-host .ui-source-icon (13 px) read --ui-icon-sm.
  3. Grid: the calc(3px - 0.125rem) switch margin is gone from the literal form. One correction to your list: deleting the rule outright left every Settings switch 0.8 px short of the row edge, and settings-edges.spec.ts failed in 4 cells. The rule is back as a plain 0.0625rem with a comment saying the spec measures it. If you would rather have no margin and a looser edge check, say so.
  4. Frames: .cl-region, .cl-ghost and .cl-chip carry a one-line "nested outer frame" comment, and UI-CONSISTENCY-AUDIT section 5 has a row.
  5. Rename field spec: asserts exactly 16 px on touch; TASTE records the iOS no-zoom exception.
  6. TASTE: the durations row says "CSS transitions".

Phone text left as is (--text-body aliases the label size): the before/after sheets of the palette and sidebar are attached below for the owner (columns: dev ink, part 2 ink, dev paper, part 2 paper; rows: palette, sidebar). Before = origin/dev 966c52e07, after = part 2 head before the switch margin fix (that change touches no text). On a phone, palette names and sidebar names go from 16 px to 13 px (unscaled rem). TASTE row 46 says phone text is "15–16 px or larger", so this needs the owner's call.

Gates (dev-02, production build):

  • scripts/verify quick on 24b7c2159, verbatim:
+---------------------------+--------------+
| Rust gate                 | Result       |
+---------------------------+--------------+
| scope guard               | PASS         |
| migration names           | PASS         |
| Nix Cargo vendoring       | PASS         |
| fmt                       | PASS         |
| doc coverage ratchet      | PASS         |
| shape lints off (D351)    | PASS         |
| doc comments on diff      | PASS         |
| fixed test waits          | PASS         |
| ui literals               | PASS         |
| tracer typecheck          | PASS         |
+---------------------------+--------------+

QUICK pre-check passed. This is not the gate: run scripts/verify fast before reporting done.
  • Budget (build.ts, transfer-budget.ts, exit 0, no cap touched): critical gzip 319,048 / 327,680; deferred gzip 58,547 / 61,440, brotli 51,906 / 55,296. tsc exit 0. Log ~/j383-e/budget.log.
  • Specs, 16-spec list: 167 passed, 2 skipped, 11 failed on the first head (~/j383-e/specs.log). 4 of the 11 were the switch margin (settings-edges), fixed; settings-edges, settings-rows, settings-rows-of-windows and replacement-alignment rerun: 44 passed (~/j383-e/specs2.log). The other 7 fail without my changes: docs-palette ×4 and toast-stack ×2 (also red on the part 1 head), palette-polish touch 393 (red on dev 966c52e07 just now, ~/j383-e/base2.log). The palette-polish 375 cell failed in the earlier runs and passed this time, so it flakes at the 54 px boundary.
  • Not run: scripts/verify fast.
**J383 part 2, review round 1 done (J401's list): `ui-consistency-d372` rebased on dev 966c52e07, head `24b7c21599ec262a647287ab3260314e2ff3ed88`** (20 commits; part 1 was already in dev). Not merged. 1. **Phone title:** `--ui-size-title: 1.375rem` in the touch block next to `--ui-size-body`; desktop stays 1.25 rem. 2. **Icons:** `.bm-icon` (18 px box) and `.ah-hero .ah-icon` read `--ui-icon`; `.bm-icon svg` (15 px) and the add-host `.ui-source-icon` (13 px) read `--ui-icon-sm`. 3. **Grid:** the `calc(3px - 0.125rem)` switch margin is gone from the literal form. **One correction to your list:** deleting the rule outright left every Settings switch 0.8 px short of the row edge, and `settings-edges.spec.ts` failed in 4 cells. The rule is back as a plain `0.0625rem` with a comment saying the spec measures it. If you would rather have no margin and a looser edge check, say so. 4. **Frames:** `.cl-region`, `.cl-ghost` and `.cl-chip` carry a one-line "nested outer frame" comment, and UI-CONSISTENCY-AUDIT section 5 has a row. 5. **Rename field spec:** asserts exactly 16 px on touch; TASTE records the iOS no-zoom exception. 6. **TASTE:** the durations row says "CSS transitions". Phone text left as is (`--text-body` aliases the label size): the **before/after sheets of the palette and sidebar are attached below** for the owner (columns: dev ink, part 2 ink, dev paper, part 2 paper; rows: palette, sidebar). Before = origin/dev 966c52e07, after = part 2 head before the switch margin fix (that change touches no text). On a phone, palette names and sidebar names go from 16 px to 13 px (unscaled rem). TASTE row 46 says phone text is "15–16 px or larger", so this needs the owner's call. **Gates (dev-02, production build):** - `scripts/verify quick` on `24b7c2159`, verbatim: ``` +---------------------------+--------------+ | Rust gate | Result | +---------------------------+--------------+ | scope guard | PASS | | migration names | PASS | | Nix Cargo vendoring | PASS | | fmt | PASS | | doc coverage ratchet | PASS | | shape lints off (D351) | PASS | | doc comments on diff | PASS | | fixed test waits | PASS | | ui literals | PASS | | tracer typecheck | PASS | +---------------------------+--------------+ QUICK pre-check passed. This is not the gate: run scripts/verify fast before reporting done. ``` - Budget (`build.ts`, `transfer-budget.ts`, exit 0, no cap touched): critical gzip 319,048 / 327,680; deferred gzip 58,547 / 61,440, brotli 51,906 / 55,296. `tsc` exit 0. Log `~/j383-e/budget.log`. - Specs, 16-spec list: 167 passed, 2 skipped, 11 failed on the first head (`~/j383-e/specs.log`). 4 of the 11 were the switch margin (settings-edges), fixed; settings-edges, settings-rows, settings-rows-of-windows and replacement-alignment rerun: 44 passed (`~/j383-e/specs2.log`). The other 7 fail without my changes: docs-palette ×4 and toast-stack ×2 (also red on the part 1 head), palette-polish touch 393 (red on dev 966c52e07 just now, `~/j383-e/base2.log`). The palette-polish 375 cell failed in the earlier runs and passed this time, so it flakes at the 54 px boundary. - Not run: `scripts/verify fast`.
phone-palette-sidebar-393x852-before-after.png
3.3 MiB
phone-palette-sidebar-375x667-before-after.png
2.9 MiB
phone-palette-sidebar-393x852-before-after.png phone-palette-sidebar-375x667-before-after.png
kayg commented 2026-11-09 06:37:42 +00:00
Author
Owner
Copy link

J383: D373 phone rows done. ui-consistency-d372 head b82460a804b9479aa244f65010a65fd6e023ae36 (21 commits on dev 411b49aa8; not merged). verify fast on compute01 is queued; its output follows in the next comment.

Change (one commit, type: phone rows read the 15 px body size…): the touch block now sets --text-body and --text-secondary to --ui-size-body (15 px) and --text-caption to --ui-size-label (13 px, the pre-change touch caption). Desktop keeps the label and eyebrow aliases. TASTE's token row records it.

Spec phone-row-text.spec.ts (production build, computed font size in px at a 16 px root): at 393×852 and 375×667 a palette result name and a session name in the sidebar are 15 px, the hint beside a palette row and a host heading in the sidebar are 13 px; at 1440×900 they are 13 / 12 / 13 / 12 px. 6 passed (~/j383-h/new-spec2.log on dev-02). The sidebar host headings (.bm-subhead) are captions, so they stay 13 px; session names are the rows.

Phone sheets attached (columns: dev ink, part 2 ink, dev paper, part 2 paper; rows: palette, sidebar, details popover; the dev column is origin/dev 966c52e07 before D372). Palette names and sidebar names are back to body size. The details popover still wraps "8 logical cores · AMD EPYC 7B13" onto two lines and is about 20 px taller, since phone body is 15 px (D372); the popover rows read the body role directly.

scripts/verify quick on this head (dev-02), ~/j383-h/quick2.log:

+---------------------------+--------------+
| scope guard               | PASS         |
| migration names           | PASS         |
| Nix Cargo vendoring       | PASS         |
| fmt                       | PASS         |
| doc coverage ratchet      | PASS         |
| shape lints off (D351)    | PASS         |
| doc comments on diff      | PASS         |
| fixed test waits          | PASS         |
| ui literals               | PASS         |
| tracer typecheck          | PASS         |
+---------------------------+--------------+

QUICK pre-check passed. This is not the gate: run scripts/verify fast before reporting done.
**J383: D373 phone rows done. `ui-consistency-d372` head `b82460a804b9479aa244f65010a65fd6e023ae36`** (21 commits on dev 411b49aa8; not merged). `verify fast` on compute01 is queued; its output follows in the next comment. **Change (one commit, `type: phone rows read the 15 px body size…`):** the touch block now sets `--text-body` and `--text-secondary` to `--ui-size-body` (15 px) and `--text-caption` to `--ui-size-label` (13 px, the pre-change touch caption). Desktop keeps the label and eyebrow aliases. TASTE's token row records it. **Spec `phone-row-text.spec.ts`** (production build, computed font size in px at a 16 px root): at 393×852 and 375×667 a palette result name and a session name in the sidebar are 15 px, the hint beside a palette row and a host heading in the sidebar are 13 px; at 1440×900 they are 13 / 12 / 13 / 12 px. 6 passed (`~/j383-h/new-spec2.log` on dev-02). The sidebar host headings (`.bm-subhead`) are captions, so they stay 13 px; session names are the rows. **Phone sheets attached** (columns: dev ink, part 2 ink, dev paper, part 2 paper; rows: palette, sidebar, details popover; the dev column is origin/dev 966c52e07 before D372). Palette names and sidebar names are back to body size. The details popover still wraps "8 logical cores · AMD EPYC 7B13" onto two lines and is about 20 px taller, since phone body is 15 px (D372); the popover rows read the body role directly. `scripts/verify quick` on this head (dev-02), `~/j383-h/quick2.log`: ``` +---------------------------+--------------+ | scope guard | PASS | | migration names | PASS | | Nix Cargo vendoring | PASS | | fmt | PASS | | doc coverage ratchet | PASS | | shape lints off (D351) | PASS | | doc comments on diff | PASS | | fixed test waits | PASS | | ui literals | PASS | | tracer typecheck | PASS | +---------------------------+--------------+ QUICK pre-check passed. This is not the gate: run scripts/verify fast before reporting done. ```
r3-phone-393x852-before-after.png
5 MiB
r3-phone-375x667-before-after.png
4.3 MiB
r3-phone-393x852-before-after.png r3-phone-375x667-before-after.png
kayg commented 2026-11-09 09:08:29 +00:00
Author
Owner
Copy link

J383: scripts/verify fast on compute01 for ui-consistency-d372 head b82460a804b9479aa244f65010a65fd6e023ae36, verbatim (log ~/j383-fast/fast.log on compute01):

+---------------------------+--------------+
| Rust gate                 | Result       |
+---------------------------+--------------+
| Nix Cargo vendoring       | PASS         |
| fmt                       | PASS         |
| migration names           | PASS         |
| migration tooling tests   | PASS         |
| VT native fmt             | PASS         |
| CLI plugin compile-out    | PASS         |
| clippy                    | PASS         |
| SSH runtime test helper   | PASS         |
| real-binary fixture paths | PASS         |
| Traefik component manifest | PASS         |
| release component publication | PASS         |
| test                      | PASS         |
| SSH inode reuse fixture   | PASS         |
| shim process-per-user     | PASS         |
| cargo-deny                | PASS         |
| gate architecture         | PASS         |
| doc coverage ratchet      | PASS         |
| shape lints off (D351)    | PASS         |
| doc comments on diff      | PASS         |
| fixed test waits          | PASS         |
| privileged integration    | SKIP         |
| probe release exclusion   | PASS         |
| nightly fuzz reporter     | PASS         |
| static builds             | PASS         |
+---------------------------+--------------+
 77253 expect() calls
Ran 50 tests across 13 files. [17.11s]
FAST_EXIT=0

One SKIP: privileged integration (the gate itself skips it on this host). The change touches no Rust. Nothing merged or deployed.

**J383: `scripts/verify fast` on compute01 for `ui-consistency-d372` head `b82460a804b9479aa244f65010a65fd6e023ae36`, verbatim (log `~/j383-fast/fast.log` on compute01):** ``` +---------------------------+--------------+ | Rust gate | Result | +---------------------------+--------------+ | Nix Cargo vendoring | PASS | | fmt | PASS | | migration names | PASS | | migration tooling tests | PASS | | VT native fmt | PASS | | CLI plugin compile-out | PASS | | clippy | PASS | | SSH runtime test helper | PASS | | real-binary fixture paths | PASS | | Traefik component manifest | PASS | | release component publication | PASS | | test | PASS | | SSH inode reuse fixture | PASS | | shim process-per-user | PASS | | cargo-deny | PASS | | gate architecture | PASS | | doc coverage ratchet | PASS | | shape lints off (D351) | PASS | | doc comments on diff | PASS | | fixed test waits | PASS | | privileged integration | SKIP | | probe release exclusion | PASS | | nightly fuzz reporter | PASS | | static builds | PASS | +---------------------------+--------------+ 77253 expect() calls Ran 50 tests across 13 files. [17.11s] FAST_EXIT=0 ``` One SKIP: privileged integration (the gate itself skips it on this host). The change touches no Rust. Nothing merged or deployed.
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
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#345
No description provided.