numpy / numpy/numpy-release

Build provenance of GCC runtime libraries in NumPy 2.5.2 macOS arm64 wheel

Open
#65 2 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Shell
Stars
4
Forks
10
Avg merge
8h 5m
Merged PRs (30d)
1

Description

We are checking the corresponding source and build provenance of compiler runtime libraries in the NumPy2.5.2 CPython3.13 macOS arm64 wheel. We have retained these public evidence joins:

  • NumPy wheel SHA256:52c808f96484f5571a5cc863775ce50247c17dfb3b0361f8ed6b4b0456f80080.
  • The publication record points to numpy/numpy-release revision72af5b09acd111d80b0e5bd9c1bcfcbec4b940b4. Its OpenBLAS requirements pin scipy-openblas32/64==0.3.34.0.0, consistent with the wheel's reported OpenBLAS version.
  • The retained OpenBLAS producer recipe selects the gcc-11.3.0-2 arm64-native provider package. That provider recipe selects versioned Conda components and removes conda-meta; the public records we found do not identify the exact underlying package builds and patches.

The three final wheel members are:

Member SHA256
numpy/.dylibs/libgfortran.5.dylib 728b8a719b0b015422632c74de2913ed14d82b319bc3a0643830bd506492662e
numpy/.dylibs/libgcc_s.1.1.dylib baee2e6118dd54c0852f881d718516d70884de84bfaa085de5d8d8f90baa3b06
numpy/.dylibs/libquadmath.0.dylib 1b341ae276c4adcb30c070b774ec6e68e819c883e897f0de8941d553f7989a0d

Could you provide the exact producer wheel/input hashes, GCC/Conda package build identities and corresponding source/patch references for these three libraries, ideally with the build or repair record that ties them to this NumPy wheel? Loader-path repair or signing can change final bytes, so an explicit before/after record would also help.

We are not reporting an exploit or evidence of tampering. The question is limited to exact source and packaging provenance. We do not need credentials, private user data or access to your infrastructure.

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start with the publication record at numpy/numpy-release revision 72af5b09acd111d80b0e5bd9c1bcfcbec4b940b4, then trace the pinned OpenBLAS requirements and retained producer/provider recipes. Compare those records with the three listed wheel-member hashes. Done means documenting exact input hashes, GCC/Conda build identities, source and patch references, and any before/after repair record tying them to the NumPy wheel.

Written by the indexing model from the issue text.

Assessment

Tech stack
python, shell
Domain
build-system, release
Issue type
Documentation
Difficulty
5/5
Estimated time
Over a week
Activity status
Active
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.