Skip to content

Conversation

@drammock
Copy link
Member

@drammock drammock commented Oct 8, 2021

following dev discussion, this should allow things like pip install mne[data] to get just the basics plus what is needed for dataset downloading (numpy, scipy, pooch, tqdm).

pyqt5>=5.10,<5.14; platform_system == "Darwin"
pyqt5>=5.10,!=5.15.2,!=5.15.3; platform_system == "Linux"
pyqt5>=5.10,!=5.15.3; platform_system != "Linux" and platform_system != "Darwin"
pyqt5>=5.10,!=5.15.3; platform_system == "Windows"
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

this is an unrelated change, but elsewhere in requirements.txt we do use platform_system == "Windows" so it seems cleaner to also use it here... unless I'm missing something?

@drammock drammock marked this pull request as draft October 8, 2021 20:11
@drammock
Copy link
Member Author

drammock commented Oct 8, 2021

ok, this is a little trickier than I thought. It would be easier if we didn't install anything from URLs in our various requirements*.txt files. Currently that means:

  • full install requirements
    • mayavi (from larsoner's fork)
  • doc requirements
    • numpydoc
    • sphinx-gallery

Personally I'd be happy to use released versions of numpydoc and sphinx-gallery instead of their dev branches... so maybe revisit this after mayavi is removed? Even then, it's not perfectly handled by the changes in this PR, because of the lines in the requirements files that specify different versions depending on platform_system. Currently that means pyqt5 and vtk; we'd have to do a little more work to parse those lines and update the extras_require entries accordingly.

so, closing for now, but happy to have input if there's a better way than what I've tried here / if some of the things I've suggested above are deal-breakers for anyone.

incidentally, I notice that the jsonschema pin from #9805 is still there, I think @larsoner forgot to remove it in #9803. I'll do that in a separate PR now.

@drammock drammock closed this Oct 8, 2021
@larsoner
Copy link
Member

larsoner commented Oct 9, 2021

We could start small and just support [data] to make it easier to use dataset fetchers

@drammock
Copy link
Member Author

drammock commented Oct 9, 2021 via email

@cbrnr
Copy link
Contributor

cbrnr commented Oct 9, 2021

We could start small and just support [data] to make it easier to use dataset fetchers

Is this a change from previous versions? Does mne.datasets require pooch now?

@agramfort
Copy link
Member

agramfort commented Oct 9, 2021 via email

@drammock drammock deleted the easier-pip-install branch October 12, 2021 21:49
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants