The suite's supabase-js clients followed redirects by default, so a 3xx
from the sandbox gateway would have carried the apikey header to another
origin. `noRedirectFetch` (`redirect: 'error'`) is passed as `global.fetch`
of both client factories; a regression test with two loopback servers
proves REST, auth and Storage calls stop at the 302. Live proof j and a
second 5/5 run are in the audit log; no production client changes.
scripts/avatar-e2e.sh read the DEV keys through `supabase status`, started
the worker on the DEV network and let the suite write and purge on DEV.
Port the V1-2 sandbox runner (`db-sandbox.sh avatar-e2e`: loopback gateway,
worker in a 600 env file) and the guarded suite (fixed URL, container and
key issuer, per-run accounts); refusals and the positive run proven under
strace in .claude/audits/2026-11-02-avatar-e2e-sandbox-only.md.
animals.ts: POST search with species filter and a probe row, read by id with
explicit columns, create with the optional first client in one call, update,
archive. relations.ts: an animal's relations in one select (voided left out),
create, end, dates, void. Kinds duplicate_relation and already_ended added.
fetchPersonalWorkspaceId, client search (keyset, one probe row), read by id
with explicit columns, create (request id), update (every field), archive,
client animals and archive preview, species labels; RecordFailure kinds of
contract §9.4 plus session_expired and network; RPC answers checked by Zod.
Unit tests and the same functions against real PostgREST on the sandbox.
A client whose only link was voided while current must not count the animal;
B linking its own animal to a client of A must not wait on A's lock. Also a
relation create racing the archive of its animal, and a replay changing the
initial relation's role.
Species rows (8 + other), animals with the D6b birth model, relations with
composite workspace keys, current/ended/voided states and a partial unique
index; definer RPCs for animals and relations (one lock order, shared write
budget, caps), invoker reads (search_animals, list_client_animals,
client_archive_preview) and search_clients with current_animal_count.
pgTAP 014 (254), parity test, sandbox suite, types from the isolated stack.
scripts/db-sandbox.sh (pnpm db:sandbox): DEV images on an own network, loopback ports distinct from DEV, no volume, secrets generated outside Git; fresh and upgrade modes through the CLI's migration up, test db and gen types with --db-url. packages/supabase/tests/sandbox: M1 sign-up and confirmation, reads and refusals per role, residual GoTrue admin path, correlation cap, concurrent sessions.
V1-1.1 (contract V1-1_OWNERS_ANIMALS.md §4, §8, §9). Tables workspaces and workspace_members, created with the profile by trigger through one atomic, idempotent write path; deferred invariant (a personal workspace has exactly its account as holder); backfill of existing profiles; access helpers caller_workspace_ids (active profile, confirmed email) and writable_workspace_ids; RLS select for members only; no privilege for anon or service_role. log_event keeps x-correlation-id only as a short token. pgTAP 012 (85 checks), census and types updated, config test.
CLAUDE.md, WORKSPACE, packages/README, README, sheet model, SUPABASE, TEST_STRATEGY,
PDF-FORMAT and ANNOTATION-RENDERING follow the deletion; RUST.md 11.23 records the census,
counts, the historical TS replay procedure and hashes; CHANGELOG, TODO, R5.7 plan and
contract.
The TS pipeline, domain, upload script and legacy scripts leave; xmldom out of the
lockfile, Dockerfile copies trimmed. LEDGER.tsv (407 cases) and FILES.tsv (112 files)
record each disposition; r4 and r5 census tests guard the tree, lockfile and
production imports.
The Rust pipeline is the official seed writer (pnpm import:equine); sheet-ref.json
and report-migration-map.json carry 12d78872...74e2 and the v1 block, the only two
seed files changed. 27c7383c...2660 stays a registered alias, honoured only through
LoadedSheet::recognises_historical_stamp (verified v1, recomputed document). M1: the
snapshot of an old report keeps its stamp; export, archive reader strict. Historical
oracles run in their pre-switch context; equivalence is crate-private.
Core rules curation::upload (manifest, local destination, bytes through the SVG reader,
plan from observed state); binary upload-sheet-assets over ureq (handshake, no overwrite,
confirm after verify, bounded retries, key never printed); pnpm assets:upload repointed,
e2e:sheet-assets for the local stack. 4 MiB cap on changes.json and references.json.
FLOOR clarification for audit prose with its negative proof.
crates/core/src/curation/draft behind the curation feature: strict changeset and references parsing as JSON.parse + Zod, the six ops, rename cascades, the B/C review rule, the R2 shape and integrity gate, migration map, document stats, seed SQL. import-equine writes both folders; --check reproduces legacy-import/ and the R5.3 draft files byte for byte (12 files, 0 differences), --out claims an absent or empty, non-symlink destination and never deletes. Parity: 825 records, 46 changes, stats, 612 leaves 442/141/29, 58 synthetic TypeScript cases (oracle r5/curation-synthetic.json).
export_pdf_for_test on the proofs build (ExportInput JSON + base64 rasters): the 14 O2
runs give the native digests through wasm; the frozen O11 scanner reads every R4 PDF and
the N-1 Zod readers accept its envelope; a stripped R4 PDF answers attachment_stripped;
the TypeScript engine reads the R4 PDFs back to the R3 outcome; E2 size ≤ R3 + 5 %. Page
resources list the images the page draws (listing every embed cost 9 MB on CAP). Node,
stub rasters, TS → Rust: demo 198 → 36 ms, L 772 → 57 ms, CAP 5.7 s → 0.22 s; files
236 KB, 386 KB, 1.6 MB against 243 KB, 463 KB, 2.2 MB. Module with the writer linked
(proofs build): 531.6 KB gzip; shipped unchanged at 426 KB until R4.6 links it.
pdf::write: text (safe_text with the frozen V8 classes, js_upper from the O4 tables,
js_trim, measure = Σ(WX || 250 + KPX) × size / 1000 over WinAnsi bytes, wrap_lines on JS
\s), afm_data generated by scripts/r4-afm-data.mjs from O5 and O4, dates (Date.UTC years
0-99 → 19xx, fr-FR / en-US long dates), input checks (branding and labels schemas,
snapshot gate with the R3 SVG reader, sheet mismatch, page caps), the header, footer,
observation and conclusion layouts into a DrawLog (per page, footers last, links by page).
cargo: O4 55,992 widths bit-exact, 1,168 wraps, safeText over every code point, upper
1,580, O6 6,468 cells + 22; O1 12 runs 2,124 ops equal at full precision; the Step 6 text
and image goldens (24) derive from the log. ts-rs: ExportInput, ExportErrorKind,
ExportWarning, LogoSlot.
Protocol v1 (task → ready → result | failed with phase and kind), the worker entry with
the glue bundled statically and the page's compiled Module posted, the file streamed into
the module's memory; client with start and task timers (paused while hidden), settle-once,
terminate always, error events never masked; module-level controller (one import at a
time, cancel on leaving Home, worker chunk prefetched once the engine is loaded); the
settlement mapped to the B.1.8 messages (OD-5: a trap, a timeout or an unreadable file is
« Erreur de lecture du PDF. », never a corrupt PDF). HomeRoute keeps its routing and size
pre-checks, loads its own index after acceptance, drops a result when the session changed.
Tests: handler and client in-process, the corpus message test (O8 51, O7 stored inputs,
O2/O3/Step 5), the six existing suites migrated to the real worker logic.
FileBytes gathers a file chunk by chunk in linear memory (capacity reserved from the
announced size, refused above the PDF cap); SheetHandle.import_pdf_bytes and
restore_copy_bytes consume it, so the worker streams file.stream() and the file lives
once, never as a JavaScript ArrayBuffer nor as a JSON-wrapped string. restore_copy:
strict UTF-8, one BOM, the OD-3 budget, draft.restore's kinds. Node: same envelopes as
import_pdf on the 51 O8 cases; rescue-copy parity with draft.restore.
Security gate finding (R4.2): prescan stopped counting once the depth exceeded 64, so a
16 MiB rescue copy nested 65 deep passed the value budget with 66 values and built its
792 MiB tree. The count now runs to the end (the depth verdict unchanged, the PDF probe
unchanged); the class is in the budget suite (48 MiB, refused) and pinned in cargo.
json_js::MAX_JSON_VALUES from the R4.2e measurements (CAP envelope 39,652 values, 612-leaf
export 23,803; ~100 B of tree per value, a 16 MiB dense document 784 MiB with the cap off,
16 MiB with it); draft.restore refuses an over-budget copy as invalid; --max-memory 1 GiB
proven to trap on the proofs build; cargo pdf_budget prints the value table, the Node
budget suite prints the per-class cost (fresh instance, W4 asserted).
locate_attachment (shared 16 MiB budget, JS-path steps in the TypeScript order, OD-3 cap
hook off until measured), pdf::import, facade import_pdf with the PdfImportKind kinds
(ts-rs exported). cargo: O7 144 inputs (locate, read, archive rows) and O8 51 cases equal
to the R3 path, 3 OD-4 differences declared (lone surrogate in an overwritten duplicate
value). Node through the shipped module: the same O8 outcomes, every reference PDF.
Module 421.7 KB gzip (W1 budget 600 KB).
Playwright harness under scripts/browser: 127.0.0.1 server with the DEPLOY
CSP, export and import baselines of the TypeScript engine under a 4 GiB RSS
watchdog (not a guaranteed memory limit), O9 browser columns in
inflate.browsers.json (26 rows agree with Node in all three engines).
Results of 2026-10-18 committed; the O9 Node test no longer expects browser
columns in inflate.json.
0279953 chose the stopped message from CoreRuntime.stopped alone: it
promised the report was kept even when the tab's storage refused the
draft (export, the banner's advice, needs the core), four surfaces still
offered a retry that could not work, and a failed recovery below the cap
was a dead end since nothing called retry().
The runtime restarts a failed recovery on the next call (compiled module
reused, cap still counts traps) and notifies status listeners. One
CoreStoppedNotice on the editor, Home, New report and the full-page
notice: reload when saved, else core.stopped_unsaved with a core-free
copy download (also on the storage-full banner) and a beforeunload
guard. Export, import and new report say the stopped message; a core
failure reading an unread draft is FAILED, not held. ESLint census of
the failure keys; core.stopped says "plusieurs erreurs".