bluesheeptoken / CPT

Compact Prediction Tree: A Lossless Model for Accurate Sequence Prediction (cython implementation)
https://cpt.readthedocs.io/en/latest/
MIT License
41 stars 7 forks source link

Wheel for Linux 64 bits #104

Closed borgeser closed 1 year ago

borgeser commented 2 years ago

Hello,

I hope this one will be easier than Mac OS. I'd like to have a wheel for Linux 64bits.

The suffix for the architecture is aarch64, here is an example of the naming : pandas-1.5.1-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl.

Thanks a lot.

borgeser commented 2 years ago

It's strange.

(I'm on Docker on my Mac M1). If I do pip install cpt, since the wheel is missing, pip is trying to build from sources and fails. However, if I clone this project and run pip wheel ., the wheel is created without problem and I can use it later.

bluesheeptoken commented 2 years ago

Hey Roland, Thanks for bringing this up.

If my memory serves me right, I don't publish wheels for arch architecture. And I think the ongoing PR might fix this (if you can try with the command I sent you for macos, I would be glad. I am struggling to find it with github mobile :D)

If that does not work, can you provide me the base image you are using ? I think it will be easier to reproduce than the macos one !

I will have a look next week once I come back from m'y hollidays :)

borgeser commented 2 years ago

Yes I think too the

[build-system]
requires ...

part is the culprit of the failure.

Have nice holidays, we'll see that next week ;)

bluesheeptoken commented 1 year ago

Indeed, github does not have arm workers :( I will soon reenable Travis CI (as numpy) because it supports ARM. I will let you know once it's done

bluesheeptoken commented 1 year ago

I am requesting travis CI some free credits to publish the wheels. I will publish wheels on pypitest as soon as I get them (if I get them). On which Python version are you running? So I can publish only this version for now.

borgeser commented 1 year ago

Thanks a lot for your dedication :) I'm using python 3.8 on my project.

bluesheeptoken commented 1 year ago

:) Here we go, project published for 3.8: pip install -i https://test.pypi.org/simple/ cpt==1.3.3.dev1 can you tell me if this working fine for you?

borgeser commented 1 year ago

:) Here we go, project published for 3.8: pip install -i https://test.pypi.org/simple/ cpt==1.3.3.dev1 can you tell me if this working fine for you?

It's working!

bluesheeptoken commented 1 year ago

Perfect, I will release that this week for all the versions :) Thanks for the feedback

bluesheeptoken commented 1 year ago

This is now completely released. Version 1.3.3 contains wheels for arm architecture 🥳