ponder-lab / Hybridize-Functions-Refactoring / #4046

90%

push

github

Leave an incomplete container's tensor-ness undetermined (#1227)

* Leave an incomplete container's tensor-ness undetermined

A container the engine reports incomplete was classified as a container of
tensors though no published fact says any element is a tensor (#1223). Such a
parameter is now left unknown, and a function whose tensor-ness rests on it
takes the existing "can't infer tensor parameters" path until wala/ML#1222.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NTnpjPjJj2aZy1ZkUD1eAn

* Keep an incomplete container unreduced beside a tensor typing

Where a parameter is already a tensor by a type hint or the tensor analysis,
the container question decides only the signature's form, and an incomplete
container is one the reduction doesn't model. Counting it there again keeps a
signature from being written from the other calls' tensors alone (#976).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NTnpjPjJj2aZy1ZkUD1eAn

* Pin a complete dict of ints as determined not to be a tensor

A dict whose keys the analysis lists and whose elements are ints isn't a
container of tensors and doesn't take the undetermined path, alone or
beside a tensor parameter: the incomplete-container rule covers only a
container that may hold elements the analysis doesn't list.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NTnpjPjJj2aZy1ZkUD1eAn

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

35 of 36 new or added lines in 3 files covered. (97.22%)

6043 of 6748 relevant lines covered (89.55%)

0.9 hits per line