Build sdformattest pybind11 library as MODULE

[traversaro]

🦟 Bug fix

Fixes segmentation fault of pyParserConfig_TEST on macOS when building the library against conda-forge dependencies.

Summary

I was investigating a segmentation fault in conda-forge/libsdformat-feedstock#134, and I isolated the problem to the sdformattest test Python module to be defined as a SHARED library. By changing it to a MODULE, the module loads fine (without segfaults) and the test passes fine.

Why the problem only occurs on conda-forge and not with Homebrew's Python (that I guess is used by tests)? I am not completely sure about this, but I guess it boils down to conda-forge using a python executable that statically links libpython, while Homebrew uses a Python that is dynamically linked to libpython, see the related issues:

• [BUG]: Segmentation Fault 11 w/ Conda + Pybind11 pybind/pybind11#3907
• Are you switching to dynamic linking? conda-forge/python-feedstock#222
• https://mail.python.org/pipermail/distutils-sig/2016-February/028275.html

Beside platform-specific problems, it seems that all other invocations of pybind11_add_module in gz projects use MODULE, so I think it make sense to switch to MODULE also here at least for consistency.

Checklist

• Signed all commits for DCO
• Added tests
• Updated documentation (as needed)
• Updated migration guide (as needed)
• Consider updating Python bindings (if the library has them)
• codecheck passed (See contributing)
• All tests passed (See test coverage)
• While waiting for a review on your PR, please help review another open pull request to support the maintainers

Note to maintainers: Remember to use Squash-Merge and edit the commit message to match the pull request summary while retaining Signed-off-by messages.

[traversaro]

I am not sure, but as this is a CMake-only change, I would guess that the Bazel failure is unrelated and actually due to some non-determinism in the Bazel build CI, also considering it also happens in #1543 .

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Please upload report for BASE (sdf15@2b583cb). Learn more about missing BASE report.

Additional details and impacted files

@@           Coverage Diff            @@
##             sdf15    #1544   +/-   ##
========================================
  Coverage         ?   92.48%           
========================================
  Files            ?      138           
  Lines            ?    18345           
  Branches         ?        0           
========================================
  Hits             ?    16967           
  Misses           ?     1378           
  Partials         ?        0           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[scpeters]

I am not sure, but as this is a CMake-only change, I would guess that the Bazel failure is unrelated and actually due to some non-determinism in the Bazel build CI, also considering it also happens in #1543 .

I think it has started failing since gazebosim/gz-math#652 was merged