release: v0.55.0 — "The scry seam"

[avrabe]

Six lanes. Version 0.54.0 → 0.55.0.

lane	what
L2	VCR-RA-010 — the scry const-remat seam. scry's singleton-interval signal promoted from a dev-dependency test to an allocator input. Eligibility is DERIVED from the emitted ARM stream; the hint only GATES consumption and can never create an eligible site. Flag-off.
L1 (v0.55 pre-merge)	--proven-safe bounds-check elision on scry's proof, fail-closed on module_sha256.
L3	VCR-DEC-001 increment 4 — the allocator models the i64 register-pair ops; 316→411 applied.
L5	Oracle steps assert EXECUTION, not exit status. 136 oracles / 295,421 emulator entries, floors measured and pinned on three surfaces.
L6	aarch64 br_table + value-carrying block/loop/if; known divergences 5 → 4.
L7	Traceability repair — rivet ours-errors 2 → 0, VCR-VER-004 roadmap entry.
#921	unmodeled-op WCET declines now name the op and byte offset (gale's schedulability input).

Gates on the assembled tree

build 0 · fmt 0 · claim 43/43 · oracle wiring 160 scripts, 0 UNDECLARED
133 test suites · frozen anchors 10/10 bit-identical
Cargo.lock refreshed (18 entries at 0.55.0)

gate-potency audit (run before tagging)

• audit 5 — skipped ≠ passed: 0 of the 9 required jobs carry a conditional or paths filter, so none can be skipped into green.
• audit 6 — missing evidence ⇒ red: 0 bare set -o pipefail; 55 steps use set -euo pipefail.
• finding → gate-potency: Version Pin Sweep covers 2 of the 4 surfaces a release bumps (Cargo.lock and npm/package.json are ungated) #924: Version Pin Sweep covers 2 of the 4 surfaces a release bumps. Cargo.lock and npm/package.json are ungated — the lock had ZERO 0.55.0 entries after the bump and is correct here only because it was checked by hand. Not a blocker for this release; filed.

Deferred to v0.56: #923 (codecov split), #924 (pin-sweep scope), #846 (gpio-thin 502→498 B).

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!