PFrames bump fix

[vadimpiven]

Greptile Summary

This PR consolidates two version-pinned whitelist entries for cupy_cuda12x (versions 13.6.0 and 14.0.1) into a single wildcard pattern cupy_cuda12x-*.whl, matching the convention already used for charset_normalizer, numba, and scipy in the same file.

• The wildcard entry matches the module set from 14.0.1; two cudnn-related modules (cupy_backends.cuda.libs.cudnn, cupyx.cudnn) that were whitelisted only in 13.6.0 are dropped — this appears intentional since 14.0.1 had already removed them.
• A patch-level changeset entry is added to trigger a release.

Confidence Score: 4/5

Safe to merge; the wildcard consolidation is consistent with patterns already used in the file and the dropped cudnn entries reflect the 14.0.1 baseline.

The change is straightforward and follows existing conventions. The only notable detail is that two cudnn-related whitelist entries present in the old 13.6.0 block are absent from the new wildcard, which is fine as long as the bump targets 14.0.1+.

checker/whitelists/python-3.12.10-rapids/linux-x64.json — confirm the cudnn module omission is intentional for the target cupy version.

Important Files Changed

Filename	Overview
.changeset/giant-monkeys-roll.md	New patch-level changeset entry for the PFrames bump fix
checker/whitelists/python-3.12.10-rapids/linux-x64.json	Consolidates two version-pinned cupy_cuda12x whitelist entries (13.6.0 and 14.0.1) into a single wildcard pattern; drops cudnn-related entries that were only present in 13.6.0

Comments Outside Diff (1)

1. checker/whitelists/python-3.12.10-rapids/linux-x64.json, line 12-23 (link)

   cudnn entries silently dropped in wildcard consolidation

   The 13.6.0 entry whitelisted cupy_backends.cuda.libs.cudnn and cupyx.cudnn (both failing with libcudnn.so.8: cannot open shared object file), but the new wildcard cupy_cuda12x-*.whl omits them. If the bumped version or a future version re-introduces cudnn as a hard import, the checker will report a new unexpected failure not covered by the whitelist. This is low-risk if the bump targets 14.0.1+ (which already lacked those entries), but worth confirming the dropped entries are intentional.

   Prompt To Fix With AI

   This is a comment left during a code review.
   Path: checker/whitelists/python-3.12.10-rapids/linux-x64.json
   Line: 12-23
   
   Comment:
   **cudnn entries silently dropped in wildcard consolidation**
   
   The 13.6.0 entry whitelisted `cupy_backends.cuda.libs.cudnn` and `cupyx.cudnn` (both failing with `libcudnn.so.8: cannot open shared object file`), but the new wildcard `cupy_cuda12x-*.whl` omits them. If the bumped version or a future version re-introduces `cudnn` as a hard import, the checker will report a new unexpected failure not covered by the whitelist. This is low-risk if the bump targets 14.0.1+ (which already lacked those entries), but worth confirming the dropped entries are intentional.
   
   How can I resolve this? If you propose a fix, please make it concise.

Prompt To Fix All With AI

Fix the following 1 code review issue. Work through them one at a time, proposing concise fixes.

---

### Issue 1 of 1
checker/whitelists/python-3.12.10-rapids/linux-x64.json:12-23
**cudnn entries silently dropped in wildcard consolidation**

The 13.6.0 entry whitelisted `cupy_backends.cuda.libs.cudnn` and `cupyx.cudnn` (both failing with `libcudnn.so.8: cannot open shared object file`), but the new wildcard `cupy_cuda12x-*.whl` omits them. If the bumped version or a future version re-introduces `cudnn` as a hard import, the checker will report a new unexpected failure not covered by the whitelist. This is low-risk if the bump targets 14.0.1+ (which already lacked those entries), but worth confirming the dropped entries are intentional.

_{Reviews (1): Last reviewed commit: "PFrames bump fix" | Re-trigger Greptile}

[gemini-code-assist]

Code Review

This pull request consolidates specific cupy_cuda12x wheel versions into a single wildcard pattern in the rapids whitelist. The review feedback correctly points out that during this consolidation, certain cudnn modules from the older wheel version were omitted, which could lead to import failures when older versions are tested. A suggestion is provided to include these missing modules in the wildcard whitelist.

[vadimpiven]

Fix for https://github.com/platforma-open/runenv-python-3/actions/runs/26435877150/job/77831333120