-
Notifications
You must be signed in to change notification settings - Fork 154
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
BLD: Add pyproject.toml #241
Conversation
I think I got every key from setup.py. The main difference is if the READTHEDOCS environment variable is set: I don't know how to change the dependencies based on that. On the other hand, given 3.8 is close to or past EOL, I suspect 3.3 is relatively uncommon, and the RTD install, without the unneeded bits, could be accomplished with pip install --no-deps.
…tes. numpy.distutils recommends setuptools<60, but that breaks pyproject.toml configuration. I like this way better. Still needs setup.py for the fortran module and for the Cheyenne configuration.
setuptools<60 just gets confused, and has no idea what the project name is. I just inverted that requirement: not sure if setuptools needs to be more recent still.
You could also deplace the requirements.txt inside the pyproject ;) |
As in, delete line 55 of pyproject.toml and move the contents of Lines 61 to 73 in cec1f70
I meant to ask earlier how important installing only mock on RTD was. The specification doesn't seem to allow conditioning requirements on environment variables, and I'm not sure TOML does either, unless there's a way for wrf-python[rtd] to have fewer dependencies than wrf-python .
|
Yes, probably you should also change |
Remove requirements.txt.
It seems someone already did that: |
The old setup script replaced the dependencies listed in requirements.txt with mock if python was old enough and nothing if python was recent. The package no longer supports python versions that old (<3.3), so installing without dependencies is likely the correct way to handle the RTD configuration.
Forgot to remove this two or three commits back.
Thanks for the PR! I was hoping to get this updated and merged yesterday, but need to need to look into why the Ubuntu CI checks are failing. |
NumPy suggested it, it might help with the distutils failure.
Let's see if that change is the one that broke things.
If it's just constants used somewhere, f2py drops the module.
Let's see if this gets Linux compiles working.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looks like the lingering failures were just a fluke w/ CI.
This all looks good to me.
Great to see the issues of this pR have been fixed. Looks good to me |
This might close #237 by specifying NumPy as a build dependency before starting the setup script.
I moved as much configuration to
pyproject.toml
as I could as an initial step to converting the build to a system available on Python 3.12.