GCC runtime libraries shipped with numpy - corresponding source
===============================================================

numpy 2.5.4's wheel (numpy-2.5.4-cp313-cp313-macosx_11_0_arm64.whl) carries three GCC runtime libraries in
numpy/.dylibs, used by its OpenBLAS:

  libgfortran.5.dylib   GPL-3.0-or-later WITH GCC-exception-3.1
  libgcc_s.1.1.dylib    GPL-3.0-or-later WITH GCC-exception-3.1
  libquadmath.0.dylib   LGPL-2.1-or-later

They are GCC 11.3.0 as conda-forge built it for osx-arm64 (package libgfortran5
11.3.0, build hdaf2cc0_26 - the gfortran-darwin-arm64-native toolchain of
https://github.com/isuruf/gcc/releases/tag/gcc-11.3.0-2, which numpy's OpenBLAS is built with).
Their code is byte-identical to that package's (checked when this pin was made).

Files:
  gcc-11.3.0.tar.xz     GNU GCC 11.3.0, https://ftp.gnu.org/gnu/gcc/gcc-11.3.0/gcc-11.3.0.tar.xz
                        (sha256 b47cf2818691f5b1e21df2bb38c795fac2cfbd640ede2d0a5e1c89e338a3ac39)
  gcc-11.3.0-arm.diff   conda-forge's Apple Silicon patch, applied to that source
  conda-forge-build.sh   conda-forge's build script (recipe/build.sh)
  conda-forge-meta.yaml  conda-forge's recipe (recipe/meta.yaml)

The three recipe files are from https://github.com/conda-forge/gfortran_impl_osx-64-feedstock
at commit 481738c977ef7b651105c9715ad2255c613500df (recipe/). The libraries' licence texts are in the app at
Contents/Resources/licenses/python-third-party/gcc-runtime/.
