ponder-lab / Hybridize-Functions-Refactoring / 25975610945

71%

main: 72%

Build Type

Pull #499

github

Committed by khatchad

Commit Message

Cache `isTensorContainer` classification on `Parameter`.

Closes #497.

`Parameter.isTensorTyped`'s Phase 3 (`hasTensorContainer`) classified a parameter as tensor-typed without populating any cache, so downstream consumers reading `Parameter.getTensorTypes()` (no-arg) saw an empty set for container parameters. `Function.inferInputSignature` (#480) silently excluded them from the inferred signature.

Add a `Boolean tensorContainer` field + `public Boolean isTensorContainer()` getter. Populate during Phase 3 of `isTensorTyped`. The new getter is the explicit query for "is this a tensor container?"; the existing `getTensorTypes()` continues to expose Ariadne's direct-tensor classification only (the asymmetry is now named on the API surface rather than hidden in the cache state).

Backward-compatible: existing callers of `isTensorTyped`/`getTensorTypes`/`getHasTensorParameter` see no behavior change. New API on `Parameter`.

Add a regression test that pins three relationships:
- Parameter-level Phase 3 classification ⇒ `Function.getHasTensorParameter()` is TRUE.
- Phase 3 cache: `Parameter.isTensorContainer()` returns TRUE.
- Asymmetry: `Parameter.getTensorTypes()` stays empty for container parameters.

Verified the test errors on `main` (compile-fail: `isTensorContainer()` undefined) and passes on this branch.

`./mvnw verify` green: 472 tests, 0 failures, 0 errors, 11 skipped (+1 from baseline).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Pull Request Pull Request #499: Cache `isTensorContainer` classification on `Parameter`

Coverage Stats

2 of 2 new or added lines in 1 file covered. (100.0%)

30 existing lines in 1 file now uncovered.

1129 of 1585 relevant lines covered (71.23%)

0.71 hits per line