Fail to install by pip on macOS 11.0.1 Big Sur about argon2-cffi HOT 11 CLOSED

In a hilarious turn of events, pip 20.3 has just been released: https://pyfound.blogspot.com/2020/11/pip-20-3-new-resolver.html

Please re-try to install with it!

Yes, it works with pip 20.3.

from argon2-cffi.

I'm afraid the bigger problem here is that pip doesn't use wheels on Big Sur yet because of the version bump to 11.0: pypa/pip#9138

It looks like an upcoming pip release should fix that and a (ugly) workaround is mentioned in the linked ticket.

As for the build problem that's gonna be a fun one, because cursory googling unveiled that blake2 currenly fails to compile on latest clangs and I don't see any fixes on neither the argon2 nor the blake repos.

from argon2-cffi.

I should note though, that I can compile it with the same clang – and supposedly CLI tools version.

Would you mind trying to work through https://github.com/hynek/argon2-cffi/blob/master/.github/CONTRIBUTING.rst#local-development-environment within a virtualenv and see if that works? There's no way in hell I'm updating to Big Sur yet, so I can't debug it myself for now.

from argon2-cffi.

In a hilarious turn of events, pip 20.3 has just been released: https://pyfound.blogspot.com/2020/11/pip-20-3-new-resolver.html

Please re-try to install with it!

from argon2-cffi.

Is there an update on this issue? I tried and failed to pip install this package. Here are my specs:
Numpy version 1.19.5
Python version 3.9.1
pip version 20.3.3
wheel version 0.36.2
macOS Big Sur 11.1 (Apple Silicon/M1)
16gb ram

from argon2-cffi.

Well, it works for others, so you're probably missing some build dependency.

from argon2-cffi.

@hynek The -msse2 flag you're using during compilation is x86-64 specific:

@stewjb has the new apple arm chip. arm CPUs don't support msse, they have their own SIMD extension called Neon

from argon2-cffi.

Actually, apple's own docs make it sound like there's no SIMD extension whatsoever in the new M1 chips: https://developer.apple.com/documentation/apple_silicon/addressing_architectural_differences_in_your_macos_code

from argon2-cffi.

We only pass that flag if optimized is True and optimized is only True on x86:

So I’m gonna need more output with recent Pythons/pips since I can’t test it myself.

from argon2-cffi.

(Also, most people use Python still under Rosetta which makes -msse2 correct.)

from argon2-cffi.

This is fixed by the outsourcing to https://github.com/hynek/argon2-cffi-bindings

Feel free to try to pip-install it and report back if it's still broken.

from argon2-cffi.