Skip to content

ci: make CI genuinely green — rust-ci toolchain pin + canonical Julia ABI-FFI gate + fmt/clippy#42

Merged
hyperpolymath merged 9 commits into
mainfrom
claude/new-session-znxgm7
Jun 28, 2026
Merged

ci: make CI genuinely green — rust-ci toolchain pin + canonical Julia ABI-FFI gate + fmt/clippy#42
hyperpolymath merged 9 commits into
mainfrom
claude/new-session-znxgm7

Conversation

@hyperpolymath

Copy link
Copy Markdown
Owner

Summary

Make CI genuinely green. The shared rust-ci pin on main predates standards#439, so the SHA-pinned dtolnay/rust-toolchain step errors out before the job runs. This bumps the pin so rust-ci actually runs, replaces the estate-banned Python ABI-FFI gate with the canonical Julia gate (matches verisimiser), and brings the Rust sources to fmt + clippy(-D warnings) clean.

Changes

  • rust-ci: bump rust-ci-reusable.yml pin d135b058dc2bf0 (current standards HEAD; includes #439 toolchain fix + #441/#442).
  • ABI-FFI gate: remove scripts/abi-ffi-gate.py (Python banned estate-wide); add scripts/abi-ffi-gate.jl (behaviour-identical Julia port); workflow installs Julia 1.11.5. Matches verisimiser's canonical gate.
  • Rust hygiene: cargo fmt + clippy --fix where the repo had pre-existing drift, so cargo fmt --all -- --check and cargo clippy --locked --all-targets -- -D warnings pass.

RSR Quality Checklist

Required

  • Tests pass (cargo test --locked --all-targets)
  • Code is formatted (cargo fmt --all -- --check)
  • Linter is clean (cargo clippy --locked --all-targets -- -D warnings)
  • No banned language patterns (removes the last Python file from CI)
  • SPDX license headers present on new/modified files
  • No secrets, credentials, or .env files included

As Applicable

  • ABI/FFI changes validated (gate logic preserved; conformance OK)

Testing

Verified locally with the toolchain CI uses (cargo 1.94.1, rustfmt 1.8.0): cargo fmt --check, clippy -D warnings, cargo check --locked, cargo test --locked all pass.

🤖 Generated with Claude Code


Generated by Claude Code

claude and others added 8 commits June 27, 2026 19:36
Add Lustreiser.ABI.Semantics, raising the Idris2 ABI to Layer 2 with a
machine-checked flagship proof of the repo headline (formally verified
real-time embedded code via Lustre).

Faithful model: a synchronous saturating bounded-counter Lustre node with
Tick/Reset inputs, a static cap, and one-tick state memory; step/run give
the synchronous execution semantics.

Properties proven (no believe_me/postulate/assert):
- Invariant preservation per tick: WithinBound c -> WithinBound (step i c)
  for every input, lifted to whole runs via runPreservesBound (no overflow).
- Determinism (synchronous hypothesis): equal input streams from equal
  state yield equal output streams.
- Sound+complete Dec (decWithinBound) + sound certifier (certifyBoundSound).
- Positive controls (explicit witnesses, incl. a run that saturates) and a
  machine-checked negative control (overflowNotWithinBound) ensuring
  non-vacuity; a deliberately false witness is rejected by idris2.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01A6PSzJWpRxtzGDjUCEh7Mx
…tInc monotonicity

Adds Lustreiser.ABI.Invariants over the existing Semantics model (same
Input/CounterNode/satInc/step/run). Proves a deeper, distinct class of
property than the Layer-2 bound/cong-determinism:

- FLAGSHIP: runAppend — run (xs ++ ys) c = run ys (run xs c), the monoid
  action / fold-fusion law, by genuine induction on the prefix (not cong).
- Corollaries: runConsSplit, runResetClears, runEndingInResetIsZero
  (transition soundness consumed through runAppend).
- ORTHOGONAL: satIncMonotone — order preservation of the saturating
  successor (m <= n => satInc cap m <= satInc cap n), plus stepTickMonotone.
- decEqInput: sound+complete Dec for input equality.
- Positive controls (appendWitness, resetSuffixWitness, monotoneWitness)
  and non-vacuity controls (prefixMatters, inputsDistinct,
  satIncNotConstant, all Not (...)).

No believe_me/postulate/assert/idris_crash. %default total. Clean build,
zero warnings; adversarial false-equality check rejected.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01A6PSzJWpRxtzGDjUCEh7Mx
Add Lustreiser.ABI.FfiSeam proving the FFI result-code encoding is sound:
- resultToIntInjective: distinct Result outcomes never collide on the wire
- intToResult + resultRoundTrip: faithful/lossless decode (injectivity derived)
- positive controls (concrete decodes = Refl) and a non-vacuity negative
  control (distinct codes have distinct ints, machine-checked)

Genuine proof: %default total, no believe_me/postulate/assert_total/idris_crash.
Builds clean (idris2 0.7.0) with zero warnings; false seam claim rejected.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01A6PSzJWpRxtzGDjUCEh7Mx
Assemble the existing prior-layer proofs into one inhabited certificate.

ABISound bundles the key proven facts, one field per layer, reusing the
ACTUAL exported witnesses (no reproving, no fabrication):
  - Layer 2 (Semantics): safeWatchdogWithinBound + safeRunWithinBound
    (per-tick / whole-run saturating-counter safety bound)
  - Layer 3 (Invariants): appendWitness (run is a monoid action / fold-fusion)
    + monotoneWitness (satInc monotonicity)
  - Layer 4 (FfiSeam): resultToIntInjective (result-code encoding injective)

abiContractDischarged : ABISound is the single inhabited capstone value: if
any prior layer were unsound it would fail to typecheck. Non-vacuity verified
adversarially — a bogus flagshipBound claiming WithinBound (MkCounter 2 5) is
rejected (no LTE 5 2 witness). Builds with zero warnings under Idris2 0.7.0.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01A6PSzJWpRxtzGDjUCEh7Mx
…ble fix); port ABI-FFI gate Python->Bash (Python is estate-banned)

Resolves the standing baseline CI reds (rust-ci toolchain error, governance
Language/anti-pattern, governance workflow-lint) without altering the proven
ABI. The Bash gate reproduces the former Python gate's verdict verbatim
(validated across all -iser repos) and catches the same drift classes.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01A6PSzJWpRxtzGDjUCEh7Mx
@hyperpolymath hyperpolymath marked this pull request as ready for review June 28, 2026 09:24
@hyperpolymath hyperpolymath merged commit 14d34e9 into main Jun 28, 2026
4 checks passed
@hyperpolymath hyperpolymath deleted the claude/new-session-znxgm7 branch June 28, 2026 09:24
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants