-
Notifications
You must be signed in to change notification settings - Fork 1.6k
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
Relax Numpy constraints #1732
Relax Numpy constraints #1732
Conversation
Only tangentially related note: the files which caused the Code format check to fail have meanwhile been fixed. |
Sounds good I have fetched upstream so if those were the only failing tests then this PR should work now. |
Hi @tburrows13! I'm not sure why you pinned Numpy to |
I think just because 1.19 was the current version when I created that PR and I didn't want to assume compatibility with unreleased versions. If they've been tested then I'm fine with increasing/removing the restriction. |
* bump numpy to 1.20 * relax numpy constraints (cherry picked from commit 7e3e8bb)
In a different repo we have been running into problems caused by moviepy's numpy version constraints. I suspected that these constraints were unnecessary so I cloned the repo and tested under different versions of numpy.
I have found that versions 1.17.3, 1.18.0, 1.20.0, 1.21.5, and 1.22.3 (latest version) all result in passing tests.
I also tested 1.17.0 which fails to install. Nevertheless it seems like all numpy versions going forward from 1.17.3 are compatible with moviepy. I would recommend that the
<=1.20.0
constraint be removed and to allow numpy versions going forward to prevent against unnecessary version conflicts for other packages.Another idea would be to add
<2.0.0
for a potentially backwards incompatible major version jump, or a more conservative<1.23.0
for minor revision changes. Patches are very safe to assume will continue to work correctly.Thanks!
PS I tested locally on a mac, I'll be checking to make sure CI here also works as expected for this change.