Repository navigation
Conversation
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Issue found by Opus 5.5 and fixed by vim and Codex 6(.1).
check_exprchecked a dummy's sort but allowed two distinctUDummysteps to match the same.mm0variable. A proof file could therefore definefasA. d1 A. d2 d1 = d2, match it against the specification'sA. d A. d d = d, and verify~ fagainst that specification.Reject a dummy already matched by an earlier unify-heap entry, skipping saved compound terms. This mirrors the kernel's existing dummy-distinctness check. Alpha-renaming, reordered and unused dummies, and repeated occurrences through
URefremain accepted.Four regression pairs in
tests/mm0_mmucover collapsed dummies, valid renaming/reordering, and saved terms interleaved with dummy matches. Each failing specification has a passing control with identical.mmuinput, so compiler rejection cannot make the negative test pass accidentally. The saved-term control also catches a scan that reads a compound term as a variable record.Validation: 65/65 C-facing MM0/MMU, MM1 and MMB tests pass with GCC. All four new regression cases and the HOL, set, Peano, peano_hex, mm0 and x86 corpora pass with GCC and Clang, including ASan/UBSan builds. The unpatched verifier accepts both new invalid specifications; the patch rejects both and accepts both positive controls. Removing the saved-term tag check makes the new positive control fail.
A broader Clang run has 7 error-reporting crashes on other malformed inputs (58/65 pass); the same crashes occur in the unpatched verifier. The mm0-js and spec-x86 suites were not run.