Skip to content

Fix checker after PFrames bump#73

Merged
vadimpiven merged 1 commit into
mainfrom
fix/pframes_bump
May 26, 2026
Merged

Fix checker after PFrames bump#73
vadimpiven merged 1 commit into
mainfrom
fix/pframes_bump

Conversation

@vadimpiven
Copy link
Copy Markdown
Contributor

@vadimpiven vadimpiven commented May 26, 2026

Greptile Summary

Updates the python-3.12.10 linux-x64 import-checker whitelist to match a PFrames package bump: cupy whitelist entries for versions 13.6.0 and 14.0.1 are replaced by a 14.1.0 entry, and new suppressions are added for librmm_cu12 25.6.0, libcudf_cu12 25.6.0, and torch 2.9.1+cpu.

  • Removes two stale cupy version blocks (13.6.0 and 14.0.1) and introduces the corresponding 14.1.0 block with the same expected shared-library errors.
  • The 14.1.0 block omits the cudnn suppressions that were present in 14.0.1; worth confirming this is intentional and that those modules now import cleanly.

Confidence Score: 4/5

Safe to merge; the change is a straightforward whitelist update with one open question about intentionally omitted cudnn suppressions.

The whitelist update is mechanical and low-risk. The only open question is whether the two cudnn module suppressions present in 14.0.1 were deliberately dropped for 14.1.0, or accidentally omitted. If they were accidentally omitted and cupy 14.1.0 still fails to load those modules, the checker would start rejecting valid environments. Everything else in the change looks correct and consistent with prior entries.

checker/whitelists/python-3.12.10/linux-x64.json — specifically the cupy 14.1.0 block and its missing cudnn entries.

Important Files Changed

Filename Overview
.changeset/tidy-sheep-see.md Adds a patch-level changeset entry for the cupy/PFrames whitelist update.
checker/whitelists/python-3.12.10/linux-x64.json Removes whitelist entries for cupy 13.6.0 and 14.0.1, adds entry for 14.1.0 (missing cudnn entries vs predecessors), and adds new entries for librmm_cu12 25.6.0, libcudf_cu12 25.6.0, and torch 2.9.1+cpu.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[PFrames bump] --> B[cupy bumped to 14.1.0]
    A --> C[librmm_cu12 25.6.0 added]
    A --> D[libcudf_cu12 25.6.0 added]
    A --> E[torch 2.9.1+cpu added]
    B --> F[Remove cupy 13.6.0 whitelist entry]
    B --> G[Remove cupy 14.0.1 whitelist entry]
    B --> H[Add cupy 14.1.0 whitelist entry]
    H --> I{cudnn suppressions?}
    I -->|Present in 14.0.1| J[cupy_backends.cuda.libs.cudnn\ncupyx.cudnn]
    I -->|Missing in 14.1.0| K[Omitted — intentional?]
Loading

Comments Outside Diff (1)

  1. checker/whitelists/python-3.12.10/linux-x64.json, line 5-16 (link)

    P2 cudnn entries absent from 14.1.0 whitelist

    The 14.0.1 (and 13.6.0) entries included two cudnn-related suppressions — cupy_backends.cuda.libs.cudnn and cupyx.cudnn — that are not present in the new 14.1.0 block. If cupy 14.1.0 genuinely resolved those missing-library errors (i.e. the modules now load cleanly without libcudnn.so.8 on the path) then the omission is correct. But if the errors still occur at runtime, the checker will flag them as unexpected failures on any environment running 14.1.0. Could you confirm whether these two modules were intentionally removed?

    Prompt To Fix With AI
    This is a comment left during a code review.
    Path: checker/whitelists/python-3.12.10/linux-x64.json
    Line: 5-16
    
    Comment:
    **cudnn entries absent from 14.1.0 whitelist**
    
    The 14.0.1 (and 13.6.0) entries included two cudnn-related suppressions — `cupy_backends.cuda.libs.cudnn` and `cupyx.cudnn` — that are not present in the new 14.1.0 block. If cupy 14.1.0 genuinely resolved those missing-library errors (i.e. the modules now load cleanly without libcudnn.so.8 on the path) then the omission is correct. But if the errors still occur at runtime, the checker will flag them as unexpected failures on any environment running 14.1.0. Could you confirm whether these two modules were intentionally removed?
    
    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/linux-x64.json:5-16
**cudnn entries absent from 14.1.0 whitelist**

The 14.0.1 (and 13.6.0) entries included two cudnn-related suppressions — `cupy_backends.cuda.libs.cudnn` and `cupyx.cudnn` — that are not present in the new 14.1.0 block. If cupy 14.1.0 genuinely resolved those missing-library errors (i.e. the modules now load cleanly without libcudnn.so.8 on the path) then the omission is correct. But if the errors still occur at runtime, the checker will flag them as unexpected failures on any environment running 14.1.0. Could you confirm whether these two modules were intentionally removed?

Reviews (1): Last reviewed commit: "Fix checker after PFrames bump" | Re-trigger Greptile

Copy link
Copy Markdown

@gemini-code-assist gemini-code-assist Bot left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Code Review

This pull request updates the cupy_cuda12x whitelist entry in the python-3.12.10 linux-x64 configuration to version 14.1.0, removing older versions. The reviewer suggested using a wildcard pattern (e.g., "cupy_cuda12x-*.whl") instead of a hardcoded version to avoid manual whitelist updates whenever the package is bumped.

Comment thread checker/whitelists/python-3.12.10/linux-x64.json Outdated
"cupyx.cutensor": "libcutensor.so.2: cannot open shared object file: No such file or directory"
},
"cupy_cuda12x-14.0.1-cp312-cp312-manylinux2014_x86_64.whl": {
"cupy_cuda12x-*.whl": {
Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@vadimpiven vadimpiven merged commit 0425cb9 into main May 26, 2026
39 of 41 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants