Support of Apple M1 ARM Chips

[pgierz]

Moin,

I was trying to follow the conda install demo, yet on new Apple computers the conda-forge channel does not provide dependencies. I'm happy to look into it if someone points me at the packaging repo.

All the best,
PG

[Chilipp]

hey @pgierz! thanks for your input! Installing psy-view without conda-forge is quite difficult. not because of psy-view itself, but rather because auf the dependencies (particularly pyqt, cartopy, numpy and pandas. They are extremely difficult to install if you want to do this yourself.

I do not yet know about the plans for conda-forge to support the M1 ARM Chips but I strongly assume that they are working on this? No idea.

I'm happy to look I to it if some points me at the packaging repo.

not sure what you mean exactly. The source for the python package is in this repository, the source for the conda-package is in https://github.com/conda-forge/psy-view-feedstock/

My apologies that I cannot be very helpful here

[pgierz]

Hey @Chilipp:

The worst one to install "by hand" (so, via pip) for me has always been cartopy. I managed this once on accident and still don't know what magic I needed in the background. Another strong point for the "take better notes" argument ;-)

For conda publishing: there normally was a separate repo that was required to upload the project to conda forge, so that's exactly the link I needed to get to look at the current details, thank you! While I have only a little bit of experience in that arena, I will try to work out the packaging details and will send you a PR with M1-ARM support if I can figure it out.

Have a nice evening!
Paul

[Chilipp]

awesome! thanks @pgierz and good luck!

[pgierz]

Well that was surprisingly straightforward. I just needed to make a PR in the Conda-forge repo: conda-forge/conda-forge-pinning-feedstock#2999

I think once it goes in that is all there is to do. I will test as soon as that happens and let you know :-)