docs(applied): rustdoc and CHANGELOG to contract statements (#365) - #367
Merged
Conversation
Terse-prose plan step 1 for catgraph-applied. Rustdoc 10710 → 7500 doc lines across 33 src files; CHANGELOG 2859 → 815 (29 version sections, one bullet per change, the 11 measured-claims figures verbatim). False claims found and fixed on the way: a `CongruenceClosure` "always returns Some" claim (three `Ok(None)` sites), a petgraph dependency that does not exist, a dead `BrauerMorphism::non_crossing` link (rustdoc + README), a `DecoratedCospan<F>` arity, a garbled braiding formula, an `f32::EPSILON` tolerance that is `1e-5`, an `i32` fixture that is `i64`. Production diff: one `if let … && …` let-chain collapse in `prop/presentation/mod.rs` (clippy `collapsible_if`, exposed when the comment between the two `if`s was deleted); no behaviour change. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Restored: the `PropSignature` f64 `Eq`/`Hash`/`Ord` implementor invariant; the `catgraph v0.11.4` pin under [0.4.0]; twelve behaviour-visible CHANGELOG sub-facts (Hypergraph method surface, `is_poisoned`, match-site `limit`/`Err` contracts, NaN buckets, rand_core public-API coupling, wasm test caveat, serde wire-format caveats, `optimize` readback check, `Color` bounds, the η known-gap test, the duplicate-port error payload, the #185 ablation count) and the overflow-class scope on the #196 bullet; `2e-6` on `E1::random`'s separation. Corrected: three retained "congruence closure decides" claims weakened to sound-and-seed-complete; `mat_kron`'s SCFM row (the Frobenius law fails, speciality holds on idempotent rigs); `transition_as_cospan` `# Panics` gains the out-of-range index arm; "four steps" in `nf_inner`; the `PetriNet` "exact inverse" wording; `Decoration::Apex` bound wording; five shipped `SfgGenerator` rigs and the `'static` bound; "In that fold" antecedent. Deleted: a vacuous colour clause on `canonical_content`; a BigInt per-operation allocation claim; a #174 history comment; a #58 test comment. Added: `parallel` forwards `catgraph/parallel` (lib.rs, README). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
`PropSignature` invariant scoped to non-NaN payloads; `transition_as_cospan` `# Panics` states the truncating `to_u64` arm; `nf_inner` "no earlier component rises"; `Checked<T>` bound spelled out and linked; `smc_nf` links restored in kb.rs; a free-standing paper fact and a why-clause deleted. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…ng (#365) Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This was referenced Aug 26, 2026
tsondru
added a commit
that referenced
this pull request
Aug 27, 2026
…ard (#298) (#369) Closes #298. `permute_side` had no non-square coverage: every pre-existing exercise is on a square matrix, so `rows != cols` never occurred and the `p.len() != expected` guard was never reached with a mismatch. Pinned on a 2 x 3 and a 3 x 2 `MatR<F64Rig>` with entries written out — the 3-cycle `rotation_left(3, 1)` on each side, and the unchanged matrix where the permutation's length matches the opposite side's arity. Both direction pins use a permutation that differs from its inverse, so each separates the braiding from its transpose. Both `matmul` results now go through `.expect` naming the guard, instead of an `if let Ok` that discarded the `Err`. The arms are unreachable: the guard makes `matmul`'s only failure condition impossible, so the issue's "reach both Err arms" cannot be satisfied by any test — a probe panicking in both arms ran 146/146 binaries green. Production diff 6 lines. Falsified against the perturbations, each red on the new test alone: guard deleted (panics at the codomain `.expect`, `expected: 3, actual: 2`); `expected` pinned to `self.cols` (domain case observes the unpermuted `[[1,2],[3,4],[5,6]]`); `expected` pinned to `self.rows` (codomain case observes the unpermuted `[[1,2,3],[4,5,6]]`); domain branch building `P` instead of `P^T` (observes `[[3,4],[5,6],[1,2]]` against `[[5,6],[1,2],[3,4]]`). Restoring `if let Ok` reddens nothing — the null that confirms the arms are dead. The issue's doc-truth rider was already discharged at #367: `mat.rs:11-12` points at `crate::mat_f64` behind the `f64-rig` feature. It is a code span rather than the intra-doc link the rider sketched, because the module is feature-gated and the link would not resolve without it. Co-authored-by: Claude Opus 5 <noreply@anthropic.com>
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
1 participant
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.
Terse-prose plan step 1 for
catgraph-applied(+ CHANGELOG). Taskmap G17-T3, tracking #365.Measured (
rg -c "^\s*(///|//!)"):src/7159 → 5350 doc lines across 33 files (whole crate 10718 → 8909);CHANGELOG.md2859 → 815 lines, 29 version sections, one bullet per change, the 11<!--m:…-->figures byte-identical.Production diff: 3 lines — one
if let … && …let-chain collapse inprop/presentation/mod.rs(clippycollapsible_if, exposed when the comment between two nestedifs was deleted; no behaviour change). Everything else is comments, rustdoc, CHANGELOG, README.False claims found in the old prose and fixed:
CongruenceClosure"always returnsSome" (threeOk(None)sites) and three sibling "decides equality" claims; apetgraphdependency that does not exist; a deadBrauerMorphism::non_crossinglink (rustdoc + README);DecoratedCospan<F>(is<Lambda, D>); a garbled braiding formula;f32::EPSILONwhere the tolerance is1e-5; ani32fixture that isi64; amat_krontable row claimingMatR's copy/add "fails speciality" (the Frobenius law fails; speciality holds on idempotent rigs); a# Panicsontransition_as_cospanmissing its out-of-range-index arm; adebug_assert!that does not exist.Reviews: four parallel executing reviews on
e2ebfa5(presentation subtree / large modules / small modules + README / CHANGELOG): 0 blocking, 6 important, 27 minor — all applied inc9fa383. Restorations the reviews forced: thePropSignaturef64Eq/Hash/Ordimplementor invariant; thecatgraph v0.11.4pin under[0.4.0]; twelve behaviour-visible CHANGELOG sub-facts (the dl-pass failure class: re-exports, error payloads, wire-format caveats);2e-6onE1::random. Delta review ofc9fa383(2 important / 5 minor: my newPropSignatureheading over-claimed on NaN;to_u64truncates, so the# Panicsarm was wrong) →ddf3207; r3 onddf3207(4 precision items on the r2 sentences:super::link paths, the exactChecked<T>bound set, an ambiguous panics sentence) →33cdeb9(links verified under--document-private-items). kb.rs's References block (BE15, Wadsley–Woods) kept as paper anchors — owner call if it should follow the deleted sentence.Gates: applied tests all features 45 binaries / 712 / 0 / 23; clippy
-D warningsdefault,serde f64-rig internal-probes,--no-default-features;RUSTDOCFLAGS=-D warnings cargo docboth lanes; doctests; fmt; workspace--nocapture146 / 2113 / 0 / 24;check_version_refs.py;check_measured_claims.py44 / 20 / 24 unchanged.Left as-is, recorded: runtime strings carrying issue numbers (
rewrite.rs:108#196,smc_nf.rs:236assert message) — behaviour, not prose;TODO(#37)comments; README:23 "overflow policy of record (#88)" prose;internal-benchfeature introduced by no CHANGELOG bullet in old or new text (pre-existing gap in the v0.1.1–v0.3.0 window the reconciliation note declares).🤖 Generated with Claude Code