Ilhan Polat

That might be true. I just did it in minpack quadpack and others and it worked independently. I think this C stuff is pretty much on consensus and no rules....

I cleaned up a bit so that only LAPACK import complaints are present and C code is not complaining. With the latest commits I get ```sh [3/114] Compiling C object...

By the way is it `blas_dep`, `lapack_dep` or `blas`, `lapack` in the meson files?

On Windows machine `scipy-openblas.pc` automatically created in the directory and `--with-scipy-openblas` takes care of what you did there. https://github.com/scipy/scipy/blob/e003d8c302c351fd78b6c4205bf6b87f70b1057f/dev.py#L514-L519 But the build still fails with the same reason. I don't...

Yes to all and the compilation intro below ```powershell C:\Users\ilhan\Documents\GitHub\scipy [rewrite_lbfgsb ≡ +1 ~1 -0 !]> git clean -xdf Removing .doit.db.dat Removing .doit.db.dir Removing build/ Removing scipy-openblas.pc Removing scipy/_build_utils/__pycache__/ Removing...

Removing it because it is not tracked by our repo. Then `python dev.py build --with-scipy-openblas` command recreates it. I checked it by cancelling the build just at the start and...

I haven't checked all failures but all give the same output that I get. Some go through the build stage though. I have to admit I am not qualified to...

So it seems like `const` is off the table with LAPACK linking (That's why we can't have nice things 😃 ). And the error message was nothing to do with...

> Builds fine, fails tests with AttributeError: module 'scipy.optimize._lbfgsb' has no attribute 'types' (f2py implementation details or leftovers?) Yes ideally all should choke on the python side import errors and...

Now it builds and passes a few tests too. Back to proper testing. EDIT: No segfaults in tests 😎