ponder-lab / Hybridize-Functions-Refactoring / #2386

81%

main: 83%

Pull #748

github

Align shape-laundering docs with the precision-favoring implementation

Reword the `shape`-read documentation on `SHAPE_MEMBER_NAME` and
`appliesNumpyToParameters`: the scan permits numpy over any shape-derived
value rather than checking static-vs-dynamic (which it cannot do here), so
say so and note the caveat (sound for the static dimensions the corpus
exercises; a genuinely dynamic dimension is missed and surfaced by the
⊤-shape reporting). Also set `valueEscapes` on the value-level numpy sink
so the shape-extractor classification stays consistent, even though a set
`sink` already dominates the decision.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01He1ULx8nRoHzGicMryry1A

84 of 85 new or added lines in 1 file covered. (98.82%)

1977 of 2438 relevant lines covered (81.09%)

0.81 hits per line