Repository navigation
Commit afeca38
committed
P-037 A1.0: port the formal kernel, controls and gates onto post-cutover main; re-measure the current record
Bootstrap only. No production code changes: the port touches no file under
frontend/, ownlang/ or rust/, which is verified against the changed-path set
rather than asserted. A1's semantics (docs/notes/p037-formal-kernel.md §8.1)
has not started.
WHAT CAME ACROSS, AND FROM WHERE
A0, A0.5 and the control authoring live on claude/p036-capability-bakeoff-fzw74t,
which also carries the P-036 comparative bakeoff. That branch is not merged and
not cherry-picked: its raw evidence is 846 files of six tools' output over a
corpus, none of which A1 reads and all of which would have arrived as unreviewed
payload behind a P-037 change. The P-037 assets are ported BY PATH onto a branch
cut fresh from main at 63148d0 (the Stage-3 merge, #359). Source tree ce2e6bc;
merge base with main 70189a3, which is also the commit the `current` control
records name.
The work those paths carry: 0355bdd (A0 kernel spike), 96bdac3 (small-width
symbolic inputs), 30fc9f0 (tractable solver harnesses), 919a908 (by-value reads
at symbolic indices), 6419b33 (collapsed()/today() by value), 9523fac (A0
terminal, 22/22 harnesses proven), 7cf1f93 (A0.5 — the G-T2a/G-T2b split),
7751d1b (the A1 sequencing ruling and the G-V4 controls), ce2e6bc (the owner
ruling on no pre-cutover ConsumesParam fix, current/post_a1 expectations, the
two-layer checker).
Deliberately left behind: docs/evidence/p036-bakeoff/raw/** and custom-codeql/,
scripts/p036_bakeoff.py, docs/notes/p036-bakeoff.md, and the 10 corpus cases
belonging to P-036 vertical B (enrollment-*, the helper/barrier/throw shapes).
8 of 18 case directories came across. The bakeoff note is still the citation for
the SHRINK hand-off, so the three places that cite it now name the branch and
ce2e6bc rather than a relative link that would dangle here — the pointer was
made honest about where it lives, the claim was not weakened to fit the port.
PHASE 0.1 — THE CURRENT RECORD, RE-MEASURED AFTER THE CUTOVER
The `current` records were measured at 70189a3, when a bare own-check.sh meant
Python. Since Stage 3 the same bare invocation means Rust, so a record whose
meaning silently changed under it had to be re-taken before A1 could rest on it.
Re-taken, not rewritten: measured_at, current and post_a1 are byte-identical to
the source branch, and each expected.json gains an append-only remeasured[]
entry (schema p037-control-expectation/2; the checker accepts /1 and /2).
Measured on both engines explicitly: all four controls UNCHANGED on both layers,
python and rust identical. --post-a1 still fails 4/4 (rc=1), as designed. The
Phase-0 STOP condition did not trigger. The engines agreeing is the two-layer
design paying out rather than luck — the fabricated `release` is emitted by the
Roslyn extractor before either engine sees a fact, which is what §8.2 read off
the facts and this now confirms by running both.
ADAPTATIONS THE CUTOVER FORCED
1. scripts/p037_controls.py passes --engine explicitly (default rust; python
measures the reference; both measures each against the same record and
reports agreement). Bare was wrong twice over post-cutover: on a machine with
no Rust candidate it exits 2 having measured nothing, and where it does run it
silently changes which engine the 70189a3 record is about. `both` agreeing is
an observation and is NOT contracted — A1 changes Rust and not Python, so a
later divergence there is a finding to classify, not one to paper over.
2. ci.yml's new formal-p037 job keeps toolchain: stable, matching rust-core
directly above it. The concrete rustc pin Stage 3 requires is a property of
action.yml — the consumer-facing distribution surface — not of this repo's own
CI, which is where a moving toolchain is supposed to break us first.
tests/test_stage3_surfaces.py reads action.yml for that check: run, green.
3. docs/proposals/README.md: only the P-037 row moved, to say something true in
this tree (Stage 3 landed, so the ruling's post-cutover condition is met). The
P-036 row is untouched — importing a dangling link into the proposals index to
carry P-036 bookkeeping through a P-037 change is not a trade worth making.
MEASURED HERE
ruff check . All checks passed!
mypy (repo config, --strict) Success: no issues found in 54 source files
tests/test_stage3_surfaces.py OK — 64 launcher invocations, 24 explicit, rc=0
formal/p037-kernel: cargo fmt --check clean
formal/p037-kernel: cargo clippy clean
formal/p037-kernel: cargo test 37 passed, 0 failed
formal/p037-kernel: 15 fast Kani harnesses 15/15 VERIFICATION:- SUCCESSFUL
scripts/p037_controls.py --engine both 4/4 all match
scripts/p037_controls.py --post-a1 MISMATCH 4/4, rc=1 (expected until A1)
Refs #304, #262.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01376o4syzCem5aQ6NssxwX21 parent 63148d0 commit afeca38
39 files changed
Lines changed: 4508 additions & 96 deletions
File tree
- .github/workflows
- corpus/p036-bakeoff
- guarded-consume-early-return
- guarded-consume-flag-branch
- guarded-consume-negation-wrapper
- guarded-consume-wrapper-forward
- gv4-control-aliased-self-null
- gv4-control-mutated-guard
- gv4-control-ref-alias-guard
- legacy-honesty-else-unresolved-forward
- docs
- notes
- proposals
- formal/p037-kernel
- src
- properties
- scripts
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
72 | 72 | | |
73 | 73 | | |
74 | 74 | | |
| 75 | + | |
| 76 | + | |
| 77 | + | |
| 78 | + | |
| 79 | + | |
| 80 | + | |
| 81 | + | |
| 82 | + | |
| 83 | + | |
| 84 | + | |
| 85 | + | |
| 86 | + | |
| 87 | + | |
| 88 | + | |
| 89 | + | |
| 90 | + | |
| 91 | + | |
| 92 | + | |
| 93 | + | |
| 94 | + | |
| 95 | + | |
| 96 | + | |
| 97 | + | |
| 98 | + | |
| 99 | + | |
| 100 | + | |
| 101 | + | |
| 102 | + | |
| 103 | + | |
| 104 | + | |
| 105 | + | |
| 106 | + | |
| 107 | + | |
| 108 | + | |
| 109 | + | |
| 110 | + | |
| 111 | + | |
| 112 | + | |
| 113 | + | |
| 114 | + | |
| 115 | + | |
| 116 | + | |
| 117 | + | |
| 118 | + | |
| 119 | + | |
| 120 | + | |
| 121 | + | |
| 122 | + | |
| 123 | + | |
| 124 | + | |
| 125 | + | |
| 126 | + | |
| 127 | + | |
| 128 | + | |
| 129 | + | |
| 130 | + | |
| 131 | + | |
| 132 | + | |
75 | 133 | | |
76 | 134 | | |
77 | 135 | | |
| |||
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
| 1 | + | |
| 2 | + | |
| 3 | + | |
| 4 | + | |
| 5 | + | |
| 6 | + | |
| 7 | + | |
| 8 | + | |
| 9 | + | |
| 10 | + | |
| 11 | + | |
| 12 | + | |
| 13 | + | |
| 14 | + | |
| 15 | + | |
| 16 | + | |
| 17 | + | |
| 18 | + | |
| 19 | + | |
| 20 | + | |
| 21 | + | |
| 22 | + | |
| 23 | + | |
| 24 | + | |
| 25 | + | |
| 26 | + | |
| 27 | + | |
| 28 | + | |
| 29 | + | |
| 30 | + | |
| 31 | + | |
| 32 | + | |
| 33 | + | |
| 34 | + | |
| 35 | + | |
| 36 | + | |
| 37 | + | |
| 38 | + | |
| 39 | + | |
| 40 | + | |
| 41 | + | |
| 42 | + | |
| 43 | + | |
| 44 | + | |
| 45 | + | |
| 46 | + | |
| 47 | + | |
| 48 | + | |
| 49 | + | |
| 50 | + | |
| 51 | + | |
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
| 1 | + | |
| 2 | + | |
| 3 | + | |
| 4 | + | |
| 5 | + | |
| 6 | + | |
| 7 | + | |
| 8 | + | |
| 9 | + | |
| 10 | + | |
| 11 | + | |
| 12 | + | |
| 13 | + | |
| 14 | + | |
| 15 | + | |
| 16 | + | |
| 17 | + | |
| 18 | + | |
| 19 | + | |
| 20 | + | |
| 21 | + | |
| 22 | + | |
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
| 1 | + | |
| 2 | + | |
| 3 | + | |
| 4 | + | |
| 5 | + | |
| 6 | + | |
| 7 | + | |
| 8 | + | |
| 9 | + | |
| 10 | + | |
| 11 | + | |
| 12 | + | |
| 13 | + | |
| 14 | + | |
| 15 | + | |
| 16 | + | |
| 17 | + | |
| 18 | + | |
| 19 | + | |
| 20 | + | |
| 21 | + | |
| 22 | + | |
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
| 1 | + | |
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
| 1 | + | |
| 2 | + | |
| 3 | + | |
| 4 | + | |
| 5 | + | |
| 6 | + | |
| 7 | + | |
| 8 | + | |
| 9 | + | |
| 10 | + | |
| 11 | + | |
| 12 | + | |
| 13 | + | |
| 14 | + | |
| 15 | + | |
| 16 | + | |
| 17 | + | |
| 18 | + | |
| 19 | + | |
| 20 | + | |
| 21 | + | |
| 22 | + | |
| 23 | + | |
| 24 | + | |
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
| 1 | + | |
| 2 | + | |
| 3 | + | |
| 4 | + | |
| 5 | + | |
| 6 | + | |
| 7 | + | |
| 8 | + | |
| 9 | + | |
| 10 | + | |
| 11 | + | |
| 12 | + | |
| 13 | + | |
| 14 | + | |
| 15 | + | |
| 16 | + | |
| 17 | + | |
| 18 | + | |
| 19 | + | |
| 20 | + | |
| 21 | + | |
| 22 | + | |
| 23 | + | |
| 24 | + | |
| 25 | + | |
| 26 | + | |
| 27 | + | |
| 28 | + | |
| 29 | + | |
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
| 1 | + | |
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
| 1 | + | |
| 2 | + | |
| 3 | + | |
| 4 | + | |
| 5 | + | |
| 6 | + | |
| 7 | + | |
| 8 | + | |
| 9 | + | |
| 10 | + | |
| 11 | + | |
| 12 | + | |
| 13 | + | |
| 14 | + | |
| 15 | + | |
| 16 | + | |
| 17 | + | |
| 18 | + | |
| 19 | + | |
| 20 | + | |
| 21 | + | |
| 22 | + | |
| 23 | + | |
| 24 | + | |
| 25 | + | |
| 26 | + | |
| 27 | + | |
| 28 | + | |
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
| 1 | + | |
| 2 | + | |
| 3 | + | |
| 4 | + | |
| 5 | + | |
| 6 | + | |
| 7 | + | |
| 8 | + | |
| 9 | + | |
| 10 | + | |
| 11 | + | |
| 12 | + | |
| 13 | + | |
| 14 | + | |
| 15 | + | |
| 16 | + | |
| 17 | + | |
| 18 | + | |
| 19 | + | |
| 20 | + | |
| 21 | + | |
| 22 | + | |
| 23 | + | |
| 24 | + | |
| 25 | + | |
| 26 | + | |
| 27 | + | |
| 28 | + | |
0 commit comments