Publish OS-specific manylinux1 wheels according to PEP 513 on PYPI #616

Activity

1. den-run-ai commented on Feb 4, 2018

   

   den-run-ai

   on Feb 4, 2018 via email

   Contributor

   More actions

   How about Mono dependency? This would require releasing wheels for every Mono release due to dependence on C-API of Mono. No one checked if pythonnet binaries are ABI compatible with various Mono versions. But historically Mono is also not very stable for pythonnet. So pythonnet had to skip few versions of Mono. Finally there are some differences in CPython compilation flags on various Linux flavors. The major one is static vs dynamic linking of libpython. So can manilinux take care of all of these? In conda version of this story, biopython just builds mono + pythonnet together.

   …

   On Sun, Feb 4, 2018, 6:50 AM Sviatoslav Sydorenko ***@***.***> wrote: Environment - Docker container, but it doesn't really matter Actual behavior Currently, if one wants to pip install the package from within linux box, they need to pre-install build dependencies. Also, pip picks sdist (tar.gz), which results into having long compilation process during install. Expected behavior I would like pip to pick the appropriate wheel and just unpack it. Extra For this to work PEP513 <https://www.python.org/dev/peps/pep-0513/> requires to have manylinux1 wheels to be built. It is completely possible to do this (incl. upload to PYPI) via Travis. I can help with this if needed. P.S. Doing same for macOS is also possible and I suspect that someone would be happy to see it as well :) — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#616>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AHgZ5Sy85Bxr2NXYXYDBLgAZkoBPWUxLks5tRaedgaJpZM4R4jpM> .
2. den-run-ai commented on Feb 5, 2018

   

   den-run-ai

   on Feb 5, 2018

   Contributor

   More actions

   closing in favor of #258, please comment with updates in there.
3. den-run-ai

   closed this as completedon Feb 5, 2018