Re-enable codegen tests on windows

[h-vetinari]

Checklist

• Used a fork of the feedstock to propose changes
• Bumped the build number (if the version is unchanged)
• Reset the build number to 0 (if the version changed)
• Re-rendered with the latest conda-smithy (Use the phrase @conda-forge-admin, please rerender in a comment in this PR for automated rerendering)
• Ensured the license file is being packaged.

Second part of replacing #12 - re-enabling the codegen tests on windows. Note this is based on #17 and only contains one extra commit on top of that.

[conda-forge-linter]

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipe) and found it was in an excellent condition.

[bstellato]

Thanks for this! I think it would be best to keep using cmake as in the other interfaces for consistency and ease of maintenance. I know it caused issues as in #10 (comment) but I hope we can solve them.

[h-vetinari]

@bstellato: Thanks for this! I think it would be best to keep using cmake as in the other interfaces for consistency and ease of maintenance. I know it caused issues as in #10 but I hope we can solve them.

Unfortunately, I believe this will not be possible. I didn't know the inner workings of the whole azure vs. make vs. MVSC, but I consider the following statement from one of the core conda folks in #12 to be definite:

@isuruf: In azure, CMake with Visual Studio generator doesn't work. You need to use NMake Makefiles or Ninja generator.

That doesn't mean it can't be solved, of course, but I believe we're gonna have to be patient for people from conda-forge who know the compiler infrastructure to help us there (or figure out what's so special about codegen vs. windows).

Incidentially, the original author of the conda-forge feedstock-recipe for osqp, @dougalsutherland, is in both @conda-forge/help-python-c and @conda-forge/core, but has been very busy recently (although sometimes seen by the odd emoji here and there ;-)).

[h-vetinari]

I can see that the log contains

%SRC_DIR%>SET "CMAKE_GENERATOR=Visual Studio 14 2015 Win64"  

just before the codegen tests fail, which seems to contradict the recommendation to use NMake. However, I can't find in the source code (nor the conda setup) where this variable is set...?

[bstellato]

Isn't it happening here? https://github.com/oxfordcontrol/osqp-python/blob/39aa8baf306b5f24a4b7d66cb0d979817a79c13f/setup.py#L24

[h-vetinari]

@bstellato
Unfortunately not, as that occurrence is already being taken care of by patch 0001...

[h-vetinari]

At long last, this now magically passed again. 🥳

[bstellato]

That's great. The mysteries of CI... Glad it is working now!