You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
docs: extend test-anchoring to the capability matrix + file residual issues
Follow-up to #622. Applies the same "claims bound to executable tests, gated
in CI" discipline to the *feature*-readiness matrix (not just the soundness
ledger), and wires in the now-tracked residuals.
- docs/CAPABILITY-MATRIX.adoc: add a "== Test anchors" section mapping each
load-bearing component (lexer, parser, type checker, effects, borrow checker,
QTT, interpreter, traits, stdlib, typed-wasm, codegen, JS-host backends, Solo
core) to the test(s) that exercise it.
- tools/check-capability-anchors.sh (wired into `just guard` + CI): fails the
build if the Test-anchors section disappears or if any test path the matrix
names goes missing — so a "works"/"enforced" row cannot outlive its test.
Verified it bites (non-zero on a missing anchor).
- docs/SOUNDNESS.adoc: wire the residual rows to their new tracking issues
(#623 interp non-tail resume, #624 Lean/Why3 return-drop) and refresh the
metatheory note to reflect the grown Coq formal/ track (#620/#621: K-1,
K-1-let, F-1).
Residuals filed as tracked issues so they aren't just prose:
- #623 — interpreter non-tail single-shot resume (shallow continuation; needs
CPS rewrite, owner-steer).
- #624 — Lean/Why3 experimental backends drop `return` control flow.
- #559 (already open) updated by comment: concrete-overlap coherence landed;
residual narrowed to generic-subsumption overlap.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01BbxKhXQwTvVgkYDgBMLJoa
0 commit comments