Closed Thoemi09 closed 2 months ago
For now, we can use setuptools as a workaround, since it ships the distutils package. In this case, we don't have to change anything in the repo.
Once we make the switch to clair (instead of cpp2py), we can check the numpy version directly in TRIQS with cmake. If we decide to only support numpy versions >= 1.8, we could merge this PR in the future.
In the second commit, I have removed
numpy_compat.py
since (as far as I know) it only checks if the numpy version is >= 1.8, which is lower than the actual triqs requirement for numpy, i.e. 1.11.