PDM, Nox and Meson

[paugier]

Oh finally it was much more changes that I expected...

This is a draft, but it gives a good idea of what it could give.

There are few interesting commands in the Makefile, in particular make should install the package in a dedicated local virtual env (.venv).

I used Nox instead of Tox because it is simpler for me.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 94.38%. Comparing base (604483b) to head (c51cb7f).
Report is 6 commits behind head on main.

❗ Current head c51cb7f differs from pull request most recent head 9f42693

Please upload reports for the commit 9f42693 to get more accurate results.

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #51      +/-   ##
==========================================
- Coverage   94.40%   94.38%   -0.03%     
==========================================
  Files           6        5       -1     
  Lines        1037     1033       -4     
==========================================
- Hits          979      975       -4     
  Misses         58       58              

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

[martibosch]

Thank you so much for doing this @paugier ! I have no experience using pdm and meson but I understand that pylandstats is now compiled using pythran in windows too, right?

Regarding nox: I am okay with switching, but if I understand this correctly, we still need to keep tox.ini? How about the tests in github actions? shouldn't tests.yml be changed to use nox?

Thank you again!

[paugier]

but I understand that pylandstats is now compiled using pythran in windows too, right?

Yes, exactly! And as a bonus there are also the files necessary to use Numba in the wheels. So Numba can be used if TRANSONIC_BACKEND is set to "numba".

How about the tests in github actions? shouldn't tests.yml be changed to use nox?

Yes, in practice, it is not necessary to have tox.ini and noxfile.py. I didn't touch the Github Actions configuration so I didn't remove tox.ini. Do you feel you can try to adapt tests.yml to use Nox? You can find an example here https://foss.heptapod.net/fluiddyn/fluidimage/-/blob/branch/default/.github/workflows/ci-linux.yml Do not hesitate to tell me if it does not work!

[martibosch]

well for some reason some pythran build errors seem to occur in github actions 😿

[paugier]

I agree that it looks reasonable. It looks a bit like a Nox bug. I'm going to have a look at that.