The large-house AC3 witness asserted an absolute 2.5 s budget for one
floor's clean-floor total. On a loaded 2-CPU machine the healthy path
took 2.7-4.3 s and the test went red while the code was fine.
The property #509 AC3 protects is structural: the summary panel builds
the space's wall masonry once and hands it to innerContourForRoom
(shared.roomGeom / shared.multiWallNodes); without it the masonry is
rebuilt for every room. Every masonry build walks the contours of all
rooms, so the test now counts reads of room.poly and compares the
floor total against one explicit spaceWallGeometry pass of the same
floor in the same run. Healthy code costs ~1.3 passes; the registered
mutant summary-area-recomputes-walls-per-room costs 21.3 and is red,
and so are the half-regressions that drop only one of the two shared
arguments (4.6 and 18.0 passes).
The count is deterministic, so machine load no longer matters.
Issue: #721
User-Visible: no
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018qZfe7YS4rqEMKoVeS3GKd
secondCardReusesPageLocale compared the count of all page requests before
and after the second card mounted. The only extra request is that card's own
plan image: under page.route the browser HTTP cache is off, so it is fetched
again, and whether it lands before the read is a race. The German locale file
itself is loaded exactly once per page. The check now counts requests for the
locale chunk only and still fails if the second card fetches it again.
Issue: #722
User-Visible: no
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018qZfe7YS4rqEMKoVeS3GKd
`workflow_dispatch` runs the file from the chosen ref, but GitHub lists a
workflow and accepts a dispatch (button, `gh workflow run`, API) only when
its file exists on the default branch. `ship-review.yml` (#696) and
`beta-derived.yml` (#697) lived only in `dev`, so neither could be started
at all, and the comment "the file runs from `--ref dev`, no mirror in
`main` needed" was wrong. Both beta steps are needed before the next
promotion would bring them to `main`.
They now follow the #623 layout instead of a full copy in `main`: a thin
caller (trigger, dispatch inputs, run-name, permission ceiling, concurrency)
calls `_ship-review.yml` / `_beta-derived.yml` at `@dev` with
`secrets: inherit`. A full copy would either need a mirror on every edit or
drift silently, and a dispatch from `main` (the button's default) would run
the stale copy; the thin caller runs the dev body from any ref. The caller
ceiling is the union of the body jobs' permissions (#556): ship-review
`contents: read` + `issues: read`, beta-derived `contents: read` +
`actions: read`; writes to `dev` stay with HP_PROCESS_TOKEN as before.
`workflow_sync` in validate.yml now compares eight files, and
test/default-branch-workflows.test.mjs lists the two dispatch-only files
explicitly with the reason checked (only `workflow_dispatch`). Workflow
tests and the #697 provenance mutant read the bodies. PROCESS.md §10.4,
§8 and §11.7 say how these are run and that a new thin file is mirrored
into `main` before it is merged into `dev`.
Issue: #716
User-Visible: no
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018qZfe7YS4rqEMKoVeS3GKd
Since #713 every raised device tile and lock badge is its floor anchor lifted
by one shared wall-top rise and room names stay on the floor, so the live
scene no longer called the #651 search. What was left of it only cost code,
build time and review attention:
- src/iso-overlays.ts: resolveIsoOverlayRigidGroups, resolveIsoOverlayCollisions
with their boundary-candidate machinery, the nudge search in
resolveIsoOverlayPlacement (the vector to the room safe point, the near-wall
test, the zoom hint), the safe point itself, the nudge/nearWall/cleared/capped
and status/reason fields, and ISO_OVERLAY_MAX_NUDGE_CSS_PX /
ISO_OVERLAY_SAFETY_GAP_CSS_PX.
- src/iso-scene-render.ts: the zoom reuse fast path and the CSS-pixel scale it
compared; a placement now depends only on anchor, owner, footprint and rise,
so zoom and stage resize reuse it by signature. residualPairs is gone and the
memo key is called layoutSignature.
- src/houseplan-card.ts: the overlay scene no longer receives the view, the
reference view or the stage rect it only fed to that scale;
data-hp-iso-nudged stays as the constant "false" that the golden
requireOneRise preflight, the live-touch smoke and the Stage 4 benchmark read.
The #585/#651 unit tests and the seven mutants that guarded only the removed
code are deleted; kept tests drop their nudge assertions, and a stage resize is
now pinned as a non-layout event. docs/ISOMETRIC.md keeps #651 as history only.
Live 2.5D output is unchanged: the 21 isometric golden scenes pass on the
accepted baselines.
Issue: #714
User-Visible: no
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018qZfe7YS4rqEMKoVeS3GKd
release.yml dispatches release-review.yml with GITHUB_TOKEN, so the run is
started by github-actions[bot], and claude-code-action refused it: "Workflow
initiated by non-human actor: github-actions (type: Bot). Add bot to
allowed_bots list" (v1.78.0: release run 36468444979, review 36468505112).
The release went out and nobody learned that the review never ran.
The review step now allows exactly github-actions[bot]. At the pinned SHA
(9cdae7f0) the action compares allowed_bots entries and the actor
case-insensitively with the `[bot]` suffix stripped, so this entry matches
GITHUB_ACTOR; any other bot is still refused, and a human dispatch never
consults the list.
independent-review no longer stops at the dispatch: it looks the run up by
workflow, branch dev, event, time and run-name "Release review <tag>" for
up to three minutes and writes the link and status to the step summary.
A run that did not appear or did not start is a warning; the release is
not blocked.
Neither file is executed from main, so no mirror is needed (§10.4).
Issue: #704
User-Visible: no
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018qZfe7YS4rqEMKoVeS3GKd
merge-candidate treated any push stderr containing "rejected" as a stale
lease. A `! [remote rejected]` from GitHub itself - in #700 the rebased
candidate changed .github/workflows/ and the conveyor token has no workflow
permission (runs 36484993494, 36487044060) - became "the branch moved after
the reviewed material (#312)", and the stderr was never printed, so the
author was sent to look for a commit that did not exist.
classifyPushRefusal now tells three outcomes apart: a stale lease
(`[rejected] (stale info)`, `fetch first`, a server-side lock race) keeps
the old behaviour; GitHub's workflow refusal (PAT, OAuth App, GitHub App,
bot and integration wordings) and any other `[remote rejected]` get their
own outcome, S6-in-progress and a comment naming the reason. The workflow
comment says what to do: the author rebases and pushes, or the owner grants
the permission. The git answer goes to the log and the comment with tokens
and credential URLs cut out; the merge-step failure comment is redacted too.
The rebase guard in _process.yml parses its push refusal with the same code
(`merge-candidate.mjs --push-refusal`): a stale lease is the old error, a
workflow refusal returns the task to S6 without review like a conflict, and
material/reuse/gate skip the rebase that never reached the branch.
Mutant push-refusal-kinds-glued restores the old regex; guard: #705 AC1.
Issue: #705
User-Visible: no
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018qZfe7YS4rqEMKoVeS3GKd
The isometric-stage3-dense-v1 runner still demanded at least one bounded
#651 nudge. Since #713 every raised device tile and lock badge is lifted by
the one shared wall-top rise and carries data-hp-iso-nudged="false", so the
Full Performance profile failed its input contract before any timing.
The contract is inverted: a single nudged raised root now fails the sample,
matching the golden requireOneRise preflight. The performance README states
the current contract, and the #570 runner-contract unit pins the new failure
text.
Issue: #719
User-Visible: no
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018qZfe7YS4rqEMKoVeS3GKd
Review r1 (High): actions/checkout passes `git fetch --no-tags` unless
`fetch-tags: true`, even with fetch-depth 0, so releaseTaggedShas() was always
empty in CI and a candidate outside the 100-run API window fell back to
event.before instead of the last release tag. Preflight and changes now fetch
tags; the workflow contract pins the option. The AC2 dev-push case now uses its
own input (dev runs only, an older `before`) instead of repeating the main call
(review r1, Low).
Issue: #703
User-Visible: no
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018qZfe7YS4rqEMKoVeS3GKd
Validate on a push to main took the range base from main's own runs only,
and skipped HEAD: the nearest judged ancestor was the previous stable, so the
whole beta line was re-judged by today's rules (run 36468413524: 55 smoke
private writes made before #629). Preflight on main used event.before, the
same old-main..candidate.
The range base now reads Validate runs of both integration branches,
counts published release tags as judged material, and accepts HEAD itself
when it already has a successful run (or a tag). A promoted SHA gets an
empty range and the dev verdict; a failed HEAD is re-judged over the same
range; a hotfix on main is judged from the candidate.
Issue: #703
User-Visible: no
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018qZfe7YS4rqEMKoVeS3GKd
The fixture repositories are removed with rmSync in each test's finally,
and that cleanup sometimes failed with ENOTEMPTY on work/.git/objects.
commit, fetch, rebase and the receiving side of push all start
`git maintenance run --auto` / `git gc --auto`; recent git (2.47+)
detaches auto maintenance by default, and a detached run creates
objects/maintenance.lock after the command has returned, i.e. while
rmSync is already walking the tree.
The environment the test already uses for core.autocrlf now also sets
maintenance.auto=false and gc.auto=0 for the working clones. The bare
origin gets receive.autogc=false, maintenance.auto=false and gc.auto=0
in its own config, since git drops GIT_CONFIG_* for the local transport's
receive-pack. The cleanup keeps rmSync in every finally (temp-dir hygiene
rule) with maxRetries/retryDelay, so a file that still appears under it
is retried instead of failing the test. No assertion changed.
Issue: #717
User-Visible: no
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018qZfe7YS4rqEMKoVeS3GKd
The second flake of smoke_dialog_polish_603: the knob slides with
`transition: left .15s` and the probe waited a fixed 220 ms, 70 ms of slack
that load ate (rightGap 3.05 and 5.6 instead of 2 in 2 of 20 loaded runs).
The probe now waits until the toggle and its pseudo-elements have no running
animation. The geometry oracle and its negative probe are unchanged.
Issue: #712
User-Visible: no
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018qZfe7YS4rqEMKoVeS3GKd
smoke_dialog_polish_603 switches the card language on every step and then
opened the room dialog with a private _openRoomEdit call followed by
updateComplete and two frames. de and fr (and the editor's settings
dictionaries) are lazy chunks: until they arrive the card's language gate
keeps the previous frame, inert and aria-busy, so the dialog is not in the
tree yet. When the chunk took longer than two frames (CI, 1 of 2 runs)
the next line read querySelector of null. Delaying the de/fr chunks by
400 ms in the harness reproduces the TypeError every time.
The step now waits until the new language is painted (no aria-busy, lang
equals the requested code), enters Plan with __hpTest.setMode, opens the
dialog with __hpTest.openRoomEdit (the real gear; the facade waits for
[data-kind="room"]) and waits until the dialog's basics card is laid out.
The covered private calls _setMode and _openRoomEdit are gone; every
check and its oracle is unchanged.
Issue: #712
User-Visible: no
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018qZfe7YS4rqEMKoVeS3GKd
untouchedDialogSaveKeepsSizes compared the stair before and after an
untouched dialog Save byte for byte, but took the reference while the
previous frame gesture's debounced save (500 ms) was still pending. That
write adopts the canonical record it sends, so under load it landed
between the two reads: the reference had x: 0.21699999999999997, the
read after Save had x: 0.217, and the check went red although the dialog
wrote nothing.
The smoke now waits for the card's own "config writes idle" condition
(no debounced save pending, no write in flight; read-only) before taking
the reference, so both sides are the same canonical stair. The exact JSON
comparison, the >1 m field and the no-history check are unchanged; a
reference that never goes idle within 5 s fails the check instead of
racing.
Issue: #708
User-Visible: no
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018qZfe7YS4rqEMKoVeS3GKd
The 6b probes entered each mode with a private _setMode call and read the
stage and paper colours 220 ms later. The mode transition interpolates
exactly those colours (inline stage background, --hp-mode-paper under
.stage.mode-transition) for its 220 ms plus a measurement frame, and it is
driven by animation frames, so under load the probe caught an
intermediate colour and plan_editor_stage_white_with_backdrop /
plan_editor_paper_white_with_backdrop went red. Slowing frames to 60 ms
reproduces both every time; three parallel copies of the smoke fail 16 of
18 runs.
Each probe now enters its mode through __hpTest.setMode and waits until
the transition has ended by the card's own markers: the stage carries
mode-<mode> and no longer mode-transition, and neither the stage nor the
paper has a running animation. The same wait precedes the View probe.
The colour assertions are unchanged; a plan stage forced to a non-white
background still fails the check.
Issue: #715
User-Visible: no
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018qZfe7YS4rqEMKoVeS3GKd
Owner decision of 2026-09-30 in #694. Switching Flat <-> 2.5D is a one-off
General settings change, and since #649 the runner measures a full config
reload for the candidate against a per-device projection flip for v1.77.0,
which alone explains most of 73.8 -> 195.7 ms. The scene build stays gated by
modelReady, firstStableRender and spaceSwitch, a UI freeze by the single
long-task ceiling; the runner still reports viewToggleMs.
Issue: #720
User-Visible: no
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018qZfe7YS4rqEMKoVeS3GKd
Since #699 the initial-View gate fails only above ceiling + band and a
decrease never fails, while the beta candidate lowers the ceiling exactly
to the fact (ratchets.mjs tighten). The #438 margin check still demanded
500 B under the ceiling and 500 B above ceiling − band, so it went red on
the first candidate with a fresh shipped bundle. It now checks the room
to the real failure edge and that a decrease stays green.
Issue: #699
User-Visible: no
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018qZfe7YS4rqEMKoVeS3GKd
Пакетное ревью задач track:ship перед бетой (PROCESS.md §11.7).
Задачи: 693. Итог: High 0 · Medium 0 · Low 1.
Опубликовано вручную по решению владельца: ship-review.yml нет в main,
и запустить его нельзя (#716). Ревью — независимый агент без контекста
реализации по промпту workflow, машинный блок — anchorBlock.
Issue: #696
User-Visible: no
The Validate artifact of run 36721827715 (c2c806fa, Linux, Chromium
151.0.7922.34) differs from the baselines only in 2.5D scenes, all expected by
AC8: the floor is no longer foreshortened, walls rise straight up, every device
tile and lock badge stands one wall-top height above its anchor, room names
keep their floor point and the home frame no longer reserves the 48 px nudge
budget. Reviewed frame by frame against the old baselines: no seam or
opening-order artefacts. 171 scenes unchanged, 130 environment witnesses.
Issue: #713
User-Visible: no
Release: v1.79.0-beta.1
Baseline-Reviewed: https://github.com/Matysh/houseplan-card/actions/runs/36721827715
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018qZfe7YS4rqEMKoVeS3GKd
The Stage 4 overlay goldens required a bounded #651 nudge; since #713 nothing
is nudged. The preflight now requires the #713 contract instead: every device
tile and lock badge is its floor anchor raised straight up by the wall-top
height, room names keep their floor point, and no root is nudged.
Issue: #713
User-Visible: no
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018qZfe7YS4rqEMKoVeS3GKd
- Vertical oblique projection: the floor matrix is the identity and a height
rises straight up by z·sin 20°, so the on-screen wall height is unchanged
and the cos 20° foreshortening of the plan, decor and anchors is gone.
- Device tiles and lock badges stand on the wall-top plane with one common
shift; the #651 placement search no longer runs in the live scene (its
removal is #714). Room names keep their Flat floor point.
- The 2.5D fit no longer reserves the 48 CSS px nudge budget.
- Switching projection keeps the camera when the previous projection was on
screen: saving the setting, entering an editor from 2.5D and adopting a warm
memo from the other projection re-read only the scalar zoom. A cold 2.5D
start still opens the 2.5D home.
- Opening faces are ordered along the oblique projector (s·y + z).
Witness: demo/smoke_iso_flat_parity.mjs (AC2–AC5, AC11) is red on the old code.
Issue: #713
User-Visible: yes
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018qZfe7YS4rqEMKoVeS3GKd
`npm run docs:accept -- --identical`: the moon lives only on the "Follow the
Sun" background at night and General settings are not in the documentation
set, so every frame matched the committed one; only the fingerprint moves.
12 reviewed frames from the golden-images artifact of Validate run
36696388011 (c8c470ed), 178 kept byte-for-byte:
- 2 new scenes (#661 AC7): day-cycle-night-moon-gibbous-dark (Moscow,
2026-10-21 18:00Z, k 0.79) and day-cycle-dusk-moon-crescent-south-dark
(Sydney, 2026-10-14 09:00Z, k 0.14) — the moon top-left behind the plan,
lit on the left, the plan covering part of the disc.
- settings-help-zoom-200-en-light and -ru-dark: the General settings card
title «Sun» became «Sun and Moon»; nothing else in the frame moved by
more than antialiasing.
- 8 isometric frames (stage6 ×5, stage3-overlays ×2, large-warm-remount):
badged device icons keep their state-free place — #711's intended
shift, merged into dev without accepting them; named here explicitly.
Release: v1.79.0-beta.1
Baseline-Reviewed: https://github.com/Matysh/houseplan-card/actions/runs/36696388011
Issue: #661
User-Visible: no
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018qZfe7YS4rqEMKoVeS3GKd
At dawn, dusk and night the environment shows the moon in the top-left
corner of the scene, behind the plan: computed in the card from the home
coordinates and the browser clock (short Meeus series + topocentric
parallax, within 1.4° / 1.8 pp of JPL Horizons), one designer image under
a continuous phase mask with the lit side always on the left (owner
2026-09-29), a feathered terminator, 3°/3 % thresholds and the 2 s fade of
the window rays. Everything but a small gate lives in the lazy
moon-runtime chunk, with its own 30 s ticker. General settings: "Sun"
becomes "Sun and Moon" with one switch, on for new installations
(DEFAULT_CONFIG), off for existing ones.
The initial View graph sat 728 B under its budget: the gate is paid for by
moving fifteen dialog-only strings of General settings into the lazy
settings dictionary (#459) and by one build fingerprint literal instead of
three, so the graph ends 4 B above dev. Golden: two new moon scenes, and
the two General settings help frames show «Sun and Moon»; the WSL artifact
test fixture now models scenes whose first capture awaits acceptance.
Issue: #661
User-Visible: yes
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018qZfe7YS4rqEMKoVeS3GKd
