Currently our release CI workflow uses python setup.py sdist bdist_wheel to create the sdist and wheel artifacts which are uploaded to PyPI during a release:

According to this article, invoking setup.py is no longer a good way to do things. In this particular case, the replacement is the build package, so we should switch to that!

How to test this?

Testing this in the pytest PR CI is not going to be possible. Just testing it locally, that the correct packages are generated, in the same place as setup.py would put them, is probably sufficient. But if you want to go the extra mile, you could create a test token on PyPI test instance and run the workflow in your fork (note: you'll need to edit this condition).