chore(memory): V1-2.2 root visibility evidence, A6 decision, BDR-089/090, LRN-102

This commit is contained in:
bchanot committed 2026-10-30 18:56:49 +01:00
1 parent a095ea0e1b
commit 958228a9de
8 files changed
+140 -9

No files matched your search

@@ -1,5 +1,5 @@
# V1-2.2 reserved ids (owner brief 2026-10-30), mutation run (evidence
# .claude/audits/2026-10-30-v1-2-2-reserved-ids-evidence.md §4): each mutant is applied to its source
# V1-2.2 reserved ids and root visibility (owner briefs 2026-10-30), mutation run (evidence
# .claude/audits/2026-10-30-v1-2-2-reserved-ids-evidence.md §4 and §7; K = reserved ids, V = root visibility): each mutant is applied to its source
# file, its cargo suite runs (`features` set when the suite needs them), the file is restored. Run
# from the repo root.
import subprocess
@@ -17,6 +17,15 @@ muts=[
(' if !append_to_first_defs(root) {',' if holds_reserved_id(root) {\n return;\n }\n if !append_to_first_defs(root) {')),
('K10 halo not written','crates/core/src/anatomy/compose.rs',' add_halo_filter(&mut asset);','','pdf_session',PDF),
('K11 reason renamed','crates/core/src/anatomy/svg/tree.rs','Self::ReservedId => "reserved_id"','Self::ReservedId => "dangling_reference"','svg_reserved_ids',''),
('V1 reader skips the root visibility check','crates/core/src/anatomy/svg/mod.rs',' sanitize::check_root_visibility(&sanitized.root)?;\n','','svg_root_visibility',''),
('V2 cleaning skips the root visibility check','crates/core/src/anatomy/svg/clean.rs','\n .and_then(|()| check_root_visibility(&read.root))','','svg_root_visibility',''),
('V3 exact case only','crates/core/src/anatomy/svg/sanitize.rs',' .eq_ignore_ascii_case("visible")',' .eq("visible")','svg_root_visibility',''),
('V4 whitespace not trimmed','crates/core/src/anatomy/svg/sanitize.rs',' .trim_matches(is_css_space)\n','','svg_root_visibility',''),
('V5 denylist of hidden only','crates/core/src/anatomy/svg/sanitize.rs',' .eq_ignore_ascii_case("visible")',' .ne("hidden")','svg_root_visibility',''),
('V6 canvas composes a hidden root','crates/core/src/anatomy/compose.rs',' if check_root_visibility(asset).is_err() {\n return Err(ComposeError::RootVisibility);\n }\n','','svg_root_visibility',''),
('V7 highlight composes a hidden root','crates/core/src/anatomy/compose.rs',' if check_root_visibility(&asset).is_err() {\n return Err(ComposeError::RootVisibility);\n }\n','','svg_root_visibility',''),
('V8 PDF gate accepts a hidden root','crates/core/src/anatomy/svg/mod.rs',' sanitize::check_root_visibility(&sanitized.root)?;\n','','pdf_session',PDF),
('V9 reason renamed','crates/core/src/anatomy/svg/tree.rs','Self::RootVisibility => "root_visibility"','Self::RootVisibility => "reserved_id"','svg_root_visibility',''),
]
for name,path,old,new,suite,features,*more in muts:
src=open(path).read()
@@ -62,3 +62,30 @@ id ») survived: no tested drawing has a root id, a mutant with no meaning for t
inherits `hidden`: 0 highlight pixels in the PDF, markers gone in the canvas too); LOW root `stroke*` inheritance onto
the point hitbox; LOW the drawing chooses which `<defs>` receives the halo (sRGB vs linearRGB, cosmetic); LOW intended
behaviour change (re-export of an archive or a logo holding the id refused).
## 7. Root visibility (brief « V1-2.2 : arbitrage définitif A6 et correction visibility », on `81f7639`)
- Red first: error names added, no check wired: `svg_root_visibility` 2 of 6 failed (reader accepted `hidden`,
composers composed it), `pdf_session` 1 failed (export accepted it); a third failure was a fixture of mine (a literal
tab is not the stored form) and the test now checks the cleaned stored form. Green once the check is wired.
- Browser proof: RUST.md §12.10 table. Composed SVG by the core at `81f7639` (stash) and with the fix, throwaway
integration test never committed; measured with Playwright Chromium 156, Firefox 157, WebKit 27.2 from the local cache
(no install, every request outside about:/blob:/data: aborted). The accepted variants' SVG are byte-identical before
and after; the three hiding roots are refused after.
- Mutations: same script, V1 to V9 added; 20 of 20 killed (K rerun, so the reserved-id protections stay effective).
- Gates on the working tree: `pnpm check:all` green (cargo 2065 passed / 0 failed summed over the feature passes, ts-rs
types unchanged, wasm, core-wasm 65, supabase 128, app 566, lint, build, e2e:ui PASSED, app image, worker sandbox
tests and runtime image; e2e:shell not run). First run stopped on an unused import in the new suite (clippy
`-D warnings`), fixed. Fixtures: cairo, Figma, Inkscape, ImageMagick and owner-folder tests green.
- Module: 560 015 → 560 377 B gzip -9 (+362 B); Node's gzip 569 817 B (budget 600 000).
- Independent reviews of the working tree (fresh, Opus). Verifier ECARTS(3), all LOW: R3 the stroke properties no
marker sets were argued harmless without proof (docs corrected, now an open finding); R7 TODO J / J6 still waited on
Illustrator (closed by BDR-090); floor guard: assertions moved into `common/sheet_readers.rs` (clarification added to
the contract). R1, R2, R4-R6, R8, R9 met. Security PASS (semgrep 352 rules, 0 findings), no regression, `osteo-halo`
protections intact; three pre-existing MEDIUM marker-hiding vectors reported for the owner: root `viewBox` framing a
view without `view_box` (all 19 equine main-view markers out of frame, horse pixel-identical, accepted Clean);
inherited `stroke-opacity` / `stroke-dasharray` (canvas outlines gone, detail trigger 0 px in the three engines);
inherited `stroke-width` on the point hitbox (13 of 19 equine markers send their click to the last hitbox in
Chromium and WebKit). LOW: halo placement in a drawing-chosen `<defs>` (Firefox, WebKit lose the halo), root
`systemLanguage` / `display` canvas-PDF divergence (whole view), colour camouflage (inherent). Not built: the class fix
changes the R3 highlight oracle and the R4 digests (owner brief: stop before such a change).
+17
View File
@@ -659,3 +659,20 @@ Also kept by owner: per-document-instance prefix required on top of per-view pre
**Rationale**: Figma writes `<mask maskUnits="userSpaceOnUse">` for frame clipping. Removal re-reads mask x/y/width/height in bounding-box units: Chromium 156 cut half stroke at (10,100), 208 px diff at ×8; mask off origin could hide content (deduced, untested). With keywords allowed: stored forms pixel-identical to raw files. Enumerated keywords carry no function, no reference.
**Alternatives**: keep removal, document 1 px loss (rejected: silent fidelity loss, worse off-origin); allow `maskUnits` only (rejected: same risk class for the two siblings); allow any value (rejected: free text where keyword suffices).
- **Status (2026-10-30)**: accepted, A6 scope part superseded by BDR-090 (Illustrator optional, 2026-10-30); mask/clip units part still in force
## BDR-089 — V1-2.2 reserved PDF halo id, refused by shared reader and composers (2026-10-30) (2026-10-30)
- **Status**: accepted
**Decision** (owner, 2026-10-30, RUST.md §12.9): PDF halo id `osteo-halo` reserved. Shared reader refuses any kept element holding it, root included, ASCII case folded (`SvgRejection::ReservedId`, `reserved_id`); cleaning, verdict, preview, loaded sheet, upload tool, PDF snapshot gate answer through same reader. Both composers refuse tree holding it (`ComposeError::ReservedId`), halo always engine's (`has_halo_filter` skip removed). Exact set `refs::RESERVED_IDS`, never `osteo-*` prefix.
**Rationale**: PDF view keeps drawing ids, engine writes `filter#osteo-halo` next to them. Defect: drawing own `filter#osteo-halo` → engine skipped its filter, transparent filter erased marker (0 highlight px Chromium 156 vs 1602); `g#osteo-halo` before first `<defs>` won resolution → halo lost (825 px). 11 historical drawings incl. DEV equine object (sha256 `0ea5202f…`) carry `osteo-overlays` → prefix reservation would refuse them. Fix: 08efd9a.
**Alternatives**: reserve `osteo-*` (rejected: refuses historical drawings); rename engine id to unguessable (rejected: changes R3 oracle + R4 digests, still collidable); strip drawing element silently (rejected: silent change, refusal explicit); composer guard only (rejected: refusal must happen before verification/publication).
## BDR-090 — V1-2.2 A6 final perimeter: Figma, Inkscape, cairo, ImageMagick; Illustrator + Affinity optional (2026-10-30) (2026-10-30)
- **Status**: accepted
**Decision** (owner, 2026-10-30, final; contract core E1, V1-2 contract §17.6): A6 tested perimeter = Figma (owner's two raw exports `owner/figma-web-20261030-{default,ids}.svg`), Inkscape 1.4.3, cairo 1.18.4, ImageMagick 7.1.2-18, all real exports read by suite. Adobe Illustrator + Affinity optional: never declared compatible/validated without real export; absence no longer blocks V1-2.2. `owner/` folder, procedure (Illustrator `presentation`/`internal-css`, Affinity `web`/`hq`), suite reading any file there kept for later. Replaces A6 part of BDR-088 (mask/clip units part still valid).
**Rationale**: owner had already lifted Illustrator requirement; earlier brief wording ("Illustrator obligatoire pour revendiquer la compatibilité Illustrator") read as A6 blocker. Compatibility claim stays evidence-bound: no export, no claim.
**Alternatives**: keep Illustrator required (rejected by owner: blocks step on tool not owned); fabricate or emulate exports (forbidden: VectorCraft not equivalent, never fake fixtures).
+1
View File
@@ -490,3 +490,4 @@ R5.4: `draft/{labels,changelog,review_report,drawing,overlay,app_review,svg_comp
- V1-2.2 owner decisions on 05de816: BDR-087 canonical stored SVG, LRN-099→LRN-100 strict reference grammar. Final review (Opus, Fable out of credits): verifier ECARTS(3) = A6 open, B9 LOW f1ef6a4 glued url( reader/composer mismatch fail-closed, FLOOR prose FP; security PASS (MEDIUM per-document prefix, 5 LOW). 05de816 = f1ef6a4 code. A6 not met. No fix, STOP at checkpoint.
- V1-2.2 finalisation 2026-10-30: Figma default Clean, ids Cleanable (g@id), pixel-identical after maskUnits/maskContentUnits/clipPathUnits allowed (BDR-088). Glued url( removed in every attribute (urls_unglued), LRN-101. check:all green 7d9ab8c. Verifier ECARTS(2): A6 Illustrator, doc drift fixed. Security PASS, pre-existing MEDIUM PDF halo id osteo-halo for owner. Not VERIFIED. STOP.
- V1-2.2 K reserved ids 2026-10-30: drawing filter#osteo-halo erased PDF marker (0 px Chromium), g#osteo-halo before defs lost halo. Reader refuses reserved_id (exact set, ASCII case), both composers fail-closed, halo always engine's. 08efd9a. check:all green, mutations 11/11, verifier ECARTS(1) LOW leaked id expansion fixed, security PASS + pre-existing MEDIUM root visibility hides markers (owner). A6 open: Illustrator.
- V1-2.2 L 2026-10-30: A6 final = Figma, Inkscape, cairo, ImageMagick; Illustrator/Affinity optional (BDR-090, BDR-088 amended). Root visibility≠visible refused (root_visibility) reader+cleaning+composers, red-first, 3 browsers, mutations 20/20, check:all green. BDR-089, LRN-102. Security PASS but 3 pre-existing MEDIUM marker-hiding vectors open (root viewBox frame, inherited stroke-opacity/dasharray, hitbox stroke-width click hijack): class fix changes R3/R4 oracles, owner decision.
+7
View File
@@ -702,3 +702,10 @@ Supersedes [[LRN-099]].
**Future application**: new SVG/CSS consumer reuses `url_openings`/`local_url_target`, never own parser. Hostile tests written apart from reader, assert browser behaviour.
Supersedes [[LRN-100]].
## LRN-102 — Unquoted heredocs expand backticks: quote the delimiter, scan before commit (2026-10-30) (2026-10-30)
**Pattern**: unquoted heredoc (`<<EOF`) runs command substitution on backticks and `$(…)` inside body. Markdown/Rust doc text with `` `id` `` became output of `id` (uid, groups incl. sudo, lxd) inside test doc comment; caught by fresh verifier before commit (V1-2.2 K, 2026-10-30). Same run: `` `tests/svg_reserved_ids.rs` `` and `` `snapshot_invalid` `` erased silently (shell errors only on stderr).
**Context**: any shell-written file holding backticks: Rust/TS doc comments, markdown, commit bodies, python snippets passed via `python3 - <<EOF`.
**Application**: always quote delimiter (`<<'EOF'`, `<<'PYEOF'`) unless expansion wanted; prefer Write/Edit tools for prose; before commit grep added lines for `uid=`, `gid=`, home paths and empty backtick pairs; treat `command not found` lines during a file write as corruption signal, re-read file.
+10 -4
View File
@@ -580,19 +580,25 @@ Order (BDR-071): P0 docs + decisions → V1-1 clients and animals → V1-2 sheet
- [x] I5 A6 inventory: Inkscape / Illustrator / Figma (version, drawing, steps, file name, folder, tests, status); Inkscape not installed → owner go 2026-10-29 to download it; inkscape.org answered 504 / timed out, so the Ubuntu archive package 1.4.3 (sha256 = signed index) + 22 libs fetched with `apt-get download`, unpacked in the scratchpad (no install, no sudo); 3 real exports + generator committed; Illustrator / Figma owner procedure; A6 stays open
- [x] I6 wasm size: contributors measured, optimisation options for later steps; cap and test kept; owner go 2026-10-29: `--remove-name-section` in scripts/wasm-build.sh (debug `name` section, 279 KB raw, −30 691 B gzip measured)
- [x] I7 gates: check:all green on 5281337 and a678faa (final rerun on the last commit); mutations 16 killed + 1 equivalent; fresh reviews (Sonnet: Fable out of credits, Opus alias 404): security BLOCK(1) HIGH url(#%61) → fixed a678faa → PASS; verifier ECARTS → LOWs fixed, A6 open; evidence §7; commit + push. Open: A6 owner exports, MEDIUM per-document prefix before any two-document page
- [ ] J finalisation and A6 Figma (brief « V1-2.2 : finalisation et intégration A6 Figma », 2026-10-30, on 98060d7)
- [x] J finalisation and A6 Figma (brief « V1-2.2 : finalisation et intégration A6 Figma », 2026-10-30, on 98060d7)
- [x] J1 Figma fixtures (`owner/figma-web-20261030-{default,ids}.svg`, owner files, never edited): both Cleanable at first; `mask@maskUnits` removal cuts 208 px (×8) in Chromium 156 → owner go 2026-10-30: allowlist `maskUnits`, `maskContentUnits`, `clipPathUnits`, values `userSpaceOnUse` | `objectBoundingBox` only; fixed expectations in `svg_clean`; pixel proof; constructs documented in exports README
- [x] J2 glued `url(` (final review LOW, f1ef6a4): the reader removes any attribute holding a `url(` that `url_openings` does not read (same helper, no second parser); composer comment fixed; tests (glued in class/result/values, valid, absent, encoded/escaped, no reference, composer vs PDF on one drawing, fail-closed branch reached)
- [x] J3 Affinity optional in A6 (owner 2026-10-30), Figma required, Illustrator expected (real Adobe Illustrator, VectorCraft not equivalent): contract, V1-2 contract §17.4, RUST.md §12, READMEs
- [x] J4 FLOOR false positive: reword `.claude/audits/2026-10-29-v1-2-2-arbitrations-security.md:225`, guard kept
- [x] J5 memory: BDR-087 checked, LRN-100 superseded after J2; per-document prefix requirement kept
- [ ] J6 gates: full cargo, svg suites, mutations, wasm + ts-rs, size < 600 000, check:all (DEV-isolated steps only), fresh independent review; commit + push; STOP (A6 open: Illustrator). Done 7d9ab8c/f976f33: check:all green, verifier ECARTS(2) (A6 Illustrator; §12.4 drift fixed after), security PASS (new pre-existing MEDIUM: PDF halo id `osteo-halo` reusable by a drawing, owner decision); left open until Illustrator
- [ ] K reserved ids and A6 Illustrator (brief « V1-2.2 : correction de sécurité PDF et poursuite A6 », 2026-10-30, on 2f1f076): fixtures and isolated stacks only, DEV untouched; no merge; STOP at checkpoint, no V1-2.3
- [x] J6 gates (A6 closed by the owner's final perimeter, BDR-090, so no longer waiting on Illustrator): full cargo, svg suites, mutations, wasm + ts-rs, size < 600 000, check:all (DEV-isolated steps only), fresh independent review; commit + push; STOP (A6 open: Illustrator). Done 7d9ab8c/f976f33: check:all green, verifier ECARTS(2) (A6 Illustrator; §12.4 drift fixed after), security PASS (new pre-existing MEDIUM: PDF halo id `osteo-halo` reusable by a drawing, owner decision); left open until Illustrator
- [x] K reserved ids and A6 Illustrator (brief « V1-2.2 : correction de sécurité PDF et poursuite A6 », 2026-10-30, on 2f1f076): fixtures and isolated stacks only, DEV untouched; no merge; STOP at checkpoint, no V1-2.3
- [x] K1 audit: `osteo-halo` written by `compose_highlight` only (PDF view keeps the drawing's ids); `has_halo_filter` skips the injection when the drawing defines `filter#osteo-halo` (hijack), a non-filter `#osteo-halo` before the first `<defs>` wins resolution; historical drawings carry `osteo-overlays` (DEV equine object included) → exact reserved set, never an `osteo-*` namespace; canvas scoped `fv.<view>.`, layer writes no id, filter `result` names local to the filter
- [x] K2 reader: `SvgRejection::ReservedId` (`reserved_id`), `refs::check_reserved_ids` (ASCII case folded, the id grammar is ASCII) in `read_svg` and in `clean_svg` after the active classification; reaches verdict, preview, loaded sheet, upload tool, PDF snapshot gate
- [x] K3 composer fail-closed: `compose_highlight` refuses a tree holding a reserved id (`ComposeError::ReservedId`), always writes its own halo (`has_halo_filter` removed); ops mapping; TS-era "never duplicated" test becomes the refusal
- [x] K4 tests: valid, filter + other elements (root included), references, adversarial spellings (case, entities, whitespace, `xml:id`, prefixed names, look-alikes), composition, PDF session (refusal + internal halo kept), equine + sha256, declared divergence for `osteo-halo-id.svg`, pixel proof in Chromium (before / after)
- [ ] K5 docs (RUST.md §12, contract §9.x hard rejections, ANNOTATION-RENDERING), Illustrator `presentation` / `internal-css` owner steps; gates (cargo, svg, mutations, wasm size, check:all DEV-isolated), fresh verifier + security on final HEAD; memory; commit + push; STOP (A6 open until Illustrator). State: check:all green, verifier ECARTS(1) LOW fixed (leaked `id` expansion in a comment), security PASS + pre-existing MEDIUM root `visibility` hides every marker (owner decision), LOW root stroke, LOW halo defs placement
- [x] K5 docs (RUST.md §12, contract §9.x hard rejections, ANNOTATION-RENDERING), Illustrator `presentation` / `internal-css` owner steps; gates (cargo, svg, mutations, wasm size, check:all DEV-isolated), fresh verifier + security on final HEAD; memory; commit + push; STOP (A6 open until Illustrator). State: check:all green, verifier ECARTS(1) LOW fixed (leaked `id` expansion in a comment), security PASS + pre-existing MEDIUM root `visibility` hides every marker (owner decision), LOW root stroke, LOW halo defs placement
- [ ] L A6 final scope and root visibility (brief « V1-2.2 : arbitrage définitif A6 et correction visibility », 2026-10-30, on 81f7639): fixtures and isolated stacks only, DEV untouched; no merge; STOP at checkpoint (VERIFIED or CHANGES REQUIRED), no V1-2.3
- [x] L1 A6 scope: Figma, Inkscape, cairo, ImageMagick = tested perimeter; Illustrator and Affinity optional, never claimed without a real export; core contract D1/B6/C1, V1-2 contract §17, RUST.md, export READMEs, TODO, BDR superseding BDR-088's A6 part; fixtures and owner procedure kept
- [x] L2 tests first (red on 81f7639): root `visibility` hides the layer (canvas, PDF), descendant `visible`, value spellings, root `style` / `<style>` forms, other root attributes; reader, cleaning, verdict, preview, canvas, PDF session, equine
- [x] L3 fix: reader refuses a root `visibility` other than `visible` (`root_visibility`), cleaning after active content, composers fail-closed; same check everywhere
- [x] L4 browser proof before / after (Chromium, Firefox, WebKit from the Playwright cache) incl. opacity, display, stroke on the root
- [ ] L5 BDR reserved ids + LRN heredoc expansion (owner go); docs; gates (cargo, fixtures, mutations, wasm + ts-rs, size, check:all), fresh verifier + security on final HEAD; commit + push; checkpoint
- [ ] V1-2.3 drawings and worker verification (M6) · V1-2.4 wireframes (editor, review, admin entry) · V1-2.5 catalogue, admin space, rights, species, MFA · V1-2.6 editor level 1 · V1-2.7 preview, review, publication UI · V1-2.8 equine working copy and review preparation · V1-2.9 FR/EN acceptance and closure (owner checkpoints V1-2.3, V1-2.4, V1-2.6, V1-2.7, V1-2.8, V1-2.9 + each DEV operation of contract §16.11)
## Release readiness track (owner 2026-10-25, distinct from the functional roadmap; nothing here authorises a release or exposure)
@@ -21,7 +21,8 @@ commit `ac71c5b` (V1-2.1 verified by the owner). Diff under review: `ac71c5b..HE
- A3 An invalid draft draws with its issues.
- A4 A document replacement redraws without re-reading the drawings.
- A5 The groups of a derived version that deletes a zone include its removal group.
- A6 Vector-tool exports cleaned and hostile files refused (see owner decision D1 below).
- A6 Vector-tool exports cleaned and hostile files refused (scope: the owner's final decision E1 below, which
replaces D1, B6 and C1 on this point).
- A7 Every edit primitive outputs a strictly parseable document, keeps existing keys and applies the
status rules (property tests).
- A8 Groups, digests and removal groups; the change list; the verdict function tested.
@@ -88,7 +89,7 @@ Brief « V1-2.2 : arbitrages du propriétaire et finalisation » on `17d9199`. D
other function still removed or refused; the cairo export keeps colours, shapes and references; fixpoint.
- B5 `sheet.edit` `remove` refused atomically (`has_dependents`, document unchanged) while `dependents` holds anything
but descendants; a removal never adds a referential issue (property test).
- B6 A6: real Inkscape 1.4.3 exports committed with their generator and provenance; Illustrator and Figma are not
- B6 (A6 scope superseded by E1) A6: real Inkscape 1.4.3 exports committed with their generator and provenance; Illustrator and Figma are not
produced or imitated; A6 stays open until the owner's files arrive.
- B7 Wasm: `--remove-name-section` only; the 600 000-byte cap and `size.test.ts` unchanged.
- B8 Still: no SQL change, no protected reader modified (A13), the N-1 copies and frozen oracle files untouched (declared
@@ -112,7 +113,7 @@ Brief « V1-2.2 : finalisation et intégration A6 Figma » after the final revie
security PASS). Delta under review: `98060d7..HEAD`. The criteria above still hold; these are added, and C1 amends D1
and B6.
- C1 A6 requires Inkscape (met, B6), Figma and Illustrator; Affinity is optional. Figma: the owner's two raw exports
- C1 (superseded by E1 for the A6 scope) A6 requires Inkscape (met, B6), Figma and Illustrator; Affinity is optional. Figma: the owner's two raw exports
`owner/figma-web-20261030-default.svg` and `-ids.svg`, never edited, committed as permanent fixtures with fixed
expectations. Illustrator: the real Adobe Illustrator only (VectorCraft is not equivalent). A6 stays open, and V1-2.2
is not VERIFIED, until the Illustrator exports are committed and pass.
@@ -125,3 +126,17 @@ and B6.
- C4 The audit sentence the floor guard read as a suppression is reworded; the guard is not disabled or waived.
- C5 Still: no SQL change, no protected reader (A13), N-1 copies, frozen oracles or equine historical data modified;
no DEV operation, no merge, no deployment, no V1-2.3.
## Final owner decisions (brief « V1-2.2 : arbitrage définitif A6 et correction visibility », 2026-10-30, on `81f7639`)
- E1 A6 scope, final: the tested perimeter is Figma (the owner's two raw exports
`owner/figma-web-20261030-{default,ids}.svg`), Inkscape 1.4.3 (three real exports and their generator), cairo
1.18.4 and ImageMagick 7.1.2-18 (real exports, `generate.sh`). Adobe Illustrator and Affinity are optional: never
declared compatible or validated without a real export read by the suite; their absence does not block V1-2.2.
The `owner/` folder, its procedure (Illustrator `presentation` / `internal-css`, Affinity `web` / `hq`) and the
suite reading any file put there stay, for a later test.
- E2 Root visibility: a root `visibility` other than `visible` is refused by the shared reader (`root_visibility`)
before a drawing can be verified or published; the cleaning, the verdict, the preview, the canvas and the PDF
snapshot gate answer it; both composers refuse such a root (fail-closed). No historical asset, oracle or golden
changes.
- E3 Still: no SQL change, no DEV operation, no historical resource modified, no merge, no deployment, no V1-2.3.
@@ -0,0 +1,49 @@
# Contract: V1-2.2 root visibility and final A6 perimeter
Owner brief « V1-2.2 : arbitrage définitif A6 et correction visibility », 2026-10-30, on `81f7639`
(branch `feature/v1-2-anatomical-sheets`). Diff under review: the working tree against `81f7639`.
## Problem
Both composers (`crates/core/src/anatomy/compose.rs`) append the engine's marker layer as the drawing
root's last child (PDF view: the drawing's own root; canvas: a root copying its attributes). A root
`visibility="hidden"` was accepted with no removal; the layer inherited it while the drawing's
children set `visible` back, so the markers vanished from the PDF and the canvas.
## Criteria
- R1 The shared reader (`read_svg`) refuses a root `visibility` other than `visible` with a closed,
stable reason `root_visibility`, before a drawing can be verified or published; the cleaning
(`clean_svg`), the verdict, the preview and its canvas, the loaded sheet, the upload tool and the PDF
snapshot gate answer through the same check (no divergent rule).
- R2 No bypass through the forms the reader accepts: letter case, CSS whitespace, character
references, CSS-wide keywords, unreadable values; equivalent styles (root `style`, root `class`,
`<style>`) never reach the root; a descendant's `visibility` cannot reach the layer.
- R3 Targeted: legitimate attributes are not refused without a demonstrated danger; other root
properties (`opacity`, `display`, `filter`, `mask`, `clip-path`, paint) are either already removed or
argued to act on drawing and markers alike, with evidence.
- R4 Both composers refuse such a root (fail-closed); the `osteo-halo` protections of `08efd9a` stay
effective.
- R5 Historical drawings, the equine fingerprint (sha256 `0ea5202f…`), frozen R3/R4 oracles, step6
goldens and N-1 copies unchanged.
- R6 Tests: reproduce the defect first, then the fix; normal drawing, hidden root, descendant
`visible`, styles, other root attributes, cleaning and canonical form, verdict, PreviewSheet, composed
canvas, real PDF export session, equine non-regression, adversarial inputs, targeted mutations,
shipped wasm module.
- R7 A6 final perimeter: Figma, Inkscape, cairo, ImageMagick (real exports, tested). Adobe Illustrator
and Affinity are OPTIONAL: their absence is NOT a gap and must not be reported as one; they must not
be declared compatible or validated anywhere without a real export; fixtures and the `owner/`
procedure stay. Contract core (D1/B6/C1 superseded by E1), V1-2 contract §17.6, RUST.md, READMEs,
TODO and the decisions registry (BDR-090, BDR-088 amended) say so consistently.
- R8 BDR (reserved `osteo-halo`, not `osteo-*`) and LRN (heredoc expansion) recorded factually.
- R9 Docs: RUST.md §12.10, V1-2 contract §9.2 and closed codes, ANNOTATION-RENDERING.
## Out of scope
No SQL, no Supabase DEV, no worker DEV, no historical resource change, no merge, no V1-2.3.
## Clarifications
- The reader-refusal assertions of `svg_reserved_ids.rs` moved, reason-parameterised and unchanged, into
`tests/common/sheet_readers.rs::assert_refused_everywhere` (shared with `svg_root_visibility.rs`); the test count is
unchanged (floor guard: assertion lines 18 → 10 in that file, by design).