Repository navigation
Move proof modules under Utf - #9
Conversation
109698c to
096c2de
Compare
zero-aska
left a comment
There was a problem hiding this comment.
Approving as a maintainer who did not write this, under MERGE-SOP.
This approval is for head 096c2de. Do not merge it before #8. This branch is stacked on #8's head ea92619, and MERGE-SOP holds a stacked PR until the branch under it lands on main. I am requesting changes on #8.
Nothing has moved since my last review. The head is still 096c2de, the base is still main 2cdbc29, and the parent is still #8's head ea92619. I checked it again anyway.
What I checked
- The top commit, "Move proof modules and spec names under Utf", is a pure rename. All its token changes are a bare
XtoUtf.X(orUtf/X). README.md, SPEC.md, bar.toml, the proofs and tests/ otherwise match #8. Utf8/, Utf16/, Main.lean, tests/ and the CI workflow are untouched. - The layout is what mission#22 decided on 2026-10-30. It has
Utf/Spec.lean,Utf/Spec/,Utf/Proofs.leanandUtf/Proofs/, and noUtf.lean. The Lake libraries areUtfSpecandUtfProofs, both indefaultTargets. comparator.json namesUtf.SpecandUtf.Proofs, and bar.toml lists the moved spec files. The existingUtf8.ProofsandUtf16.Proofsnamespaces stay as they were. lake buildpasses on this head with 37 jobs. The only warnings are the 35 expectedsorryones in the challenge file.bar/check.py --offline --run-testson this head:- Item 1 is OPEN, since no approval is recorded. The spec hash is now 71efe843.... No approval existed, so nothing is voided.
- Items 2 to 6 PASS. Item 2: "What is proved" names all 35 theorems by their full
Utf.Specname. Item 3:lake comparator --paranoidaccepts all 35 under every bundled kernel. Item 5: 2,818,540 inputs, 0 failures. - Item 7 SKIP (offline).
- Statement check against main: I dumped the full term text of every type and definition in the spec modules, on main and on this head. With
Utf.Specmapped back toSpec, all 120 constants match. The one exception is #8's new conjunct insubstitution_vectors. Without that conjunct, the type is exactly main's. An earlier dump of all 903 solution constants against #8 also matched, with the same proof hashes and#print axioms. - There are two
bar / barruns on 096c2de, both success.git merge-treeof this head onto main is clean.
About #8
This branch also carries #8's "Mutation evidence limits" section in SPEC.md, through #8's commit. The top commit does not touch it, so the fix belongs in #8.
- Suppose #8 is fixed with a new commit and merged with a merge commit. Then this head still merges cleanly, and the section is gone from the result. A docstring fix in #8's
Spec.leanalso lands inUtf/Spec.lean. I simulated both withgit merge-tree. This approval then stays on the current head. - If #8 is amended, squashed or rebased, rebase this branch onto it. That makes a new head, which needs a fresh approval.
- If #8 brings in
tests/planted_bugs.pyfromclaude/review-followups, that script runslake build Proofs. On this head that gives "unknown targetProofs", so this PR would need to change it tolake build UtfProofs, with a fresh approval.
Non-blocking
- Three comments in the spec files still give the old paths:
Utf/Spec.leanlines 12 and 14 (Spec/Defs.lean,Spec/Utf16.lean) andUtf/Spec/Utf16.leanline 150 (Spec/Defs.lean). Everywhere else the PR writesUtf/Spec/.... These files count toward the approval hash, so it is cheaper to fix them before a spec approval is recorded. If you rebase onto a changed #8, fix them in the same push.
Generated by Claude Code
Resolve Utf.Spec.substitution_vectors docstring; point planted_bugs.py at UtfProofs. Co-authored-by: ahraz <ahrazzle@users.noreply.github.com>
Co-authored-by: ahraz <ahrazzle@users.noreply.github.com>
The tree is unchanged. This refreshes the merge-base so a last-push approval can apply.
zero-aska
left a comment
There was a problem hiding this comment.
Verified at 74b9be6 on the box. Head is now fc1f3d7, a tree-identical merge of main after #8 (tree 2e3111a65eec8d378c69576d1b260feaaad49d75 is unchanged).
This PR merges #8 and moves the spec and proof modules under Utf. Every theorem statement is identical to #8's, ignoring namespace. Utf.Spec.substitution_vectors matches Spec.substitution_vectors, including the added conjunct.
- lake build passed, including the UtfSpec and UtfProofs libraries. tests/run_tests.py: 8,714,312/8,714,312 requests, 0 failures. tests/cross_check.py: 2,358,234 byte strings OK. tests/planted_bugs.py --count 20000: 28 of 30 planted bugs caught; the two at 0 failures ("Leads 0xF5 to 0xF7 start four-byte sequences" and "A bad last byte given its own branch") are marked equivalent mutations, and lake build UtfProofs still fails for both. The generated record matches the committed PLANTED-BUGS.md. mission bar/check.py --offline --run-tests at current mission main: in-progress, no failures; item 1 remains OPEN. CI bar/bar was green on 74b9be6.
- No theorem statement is weakened. No new sorry, axiom, native_decide, implemented_by, extern, or unsafe outside the spec statement files.
Approving the module move. This is not a specification approval.
Move specification and proof modules beneath
Utfand update imports, Lake roots, configuration and documentation to match. Prefix shared specification declarations withUtf.Spec.Part of Proven-Commons/mission#22. The other libraries have separate adoption PRs.
Merge order: this lands second, after #8. The branch is rebased on top of #8's branch (base stays
main), so until #8 merges this PR also shows #8's commit; review only the top commit, "Move proof modules and spec names under Utf". Once #8 merges, the diff here is the move alone. In the rebase, #8's README/SPEC row ("three cut-short sequences") and its "Mutation evidence limits" section were carried into the moved docs; the new vector moved withUtf/Spec.leanandUtf/Proofs/Main.lean.Validation, on the rebased head:
lake buildpasses, including the relocated proof modules.python3 tests/run_tests.py: 8714312/8714312 requests agree, 0 failures (record unchanged).🤖 Generated with Claude Code
https://claude.ai/code/session_01QwrvNdsMA6RtgKxXn1Pmhf