New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Including bottleneck as dependency in install_requires in setup.py causes installation to fail #195
Comments
Are there any plans to fix this? This continues to affect multiple developers on my team since we have an indirect dependency on this package. |
Can you have your team test your proposed solution with and without numpy preinstalled? If it work make a PR |
I proposed a workaround for downstream packages, but not a solution. Basically it's not safe to import I think the right long-term solution here is probably PEP 518. It requires downstream users to have |
All this packaging stuff is a mystery to me. You ask if I have plans to fix it but you don't have a solution? Is that the status? |
If every bug report came with a solution for free, my job would be a lot easier. If you don't care to address this, feel free to close. I simply asked about the status since there had been no reply to my original report. |
I wasn't expecting a solution for free. I was just confused about the status---whether or not there was a proposed solution. |
Sorry, I misunderstood 🤐. I can try to open a PR soon that adds a To be fair, Python packaging is a mess. It seems like PEP 518 offers a sane way forward, but unfortunately it's going to be awhile before it's safe to assume that everyone has a sufficiently recent version of The other workaround solution (which would be a pain) is to make your |
So if you added |
Yes, that's how I understand it. |
Just for the record, the ideal solution is indeed the |
The issue reported here is addressed by #203. However, I realized that the specific case I described in this issue (using |
Here's a minimal test case. Use the
setup.py
below and then runpython setup.py install
orpython setup.py develop
in a clean environment (i.e. one that doesn't already havenumpy
installed):The full failure is listed below. A workaround is to add
numpy
tosetup_requires
, but it seems like downstream packages should not be required to do this. This is probably not noticed by most users sincenumpy
is already installed in most environments, but the current behavior is a bug.The text was updated successfully, but these errors were encountered: