Skip to content
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

conda distribution #160

Open
analog-cbarber opened this issue Jul 12, 2022 · 4 comments
Open

conda distribution #160

analog-cbarber opened this issue Jul 12, 2022 · 4 comments

Comments

@analog-cbarber
Copy link

Would be great if this was available as a conda package (e.g. on conda-forge).

@BrianSong
Copy link
Collaborator

Hi @analog-cbarber, thanks for bringing this up. The conda distribution sounds good to have - even though it might not be high priority, I can talk to our release engineer to see if we can make this happen.

Btw, is there any problem with current PyPI support?

@analog-cbarber
Copy link
Author

I am not aware of problems with the pypi packages, but there are a number of problems including a non-conda package into a conda ecosystem. Although you can install packages into a conda environment using pip, there is no way for a conda package to declare a dependency on a pip package, so any downstream packages that use this will either need to implement some sort of post-install hack or force the user to do a separate manual install of any required pip packages.

Another concern with pip packages is that pip will happily overwrite packages installed by conda with incompatible versions. This doesn't happen that often and is usually not a problem for pure python packages, but for those that have binary dependencies it can lead to crashes.

Since the ML community does use conda environments quite frequently, it would be very welcome for there to be a conda distribution as there is for other commonly used ML packages such as tensorflow and pytorch.

It probably is a non-trivial amount of work for your release engineer, but shouldn't require much energy to maintain once you have set it up, so hopefully you can get around to it eventually.

@BrianSong
Copy link
Collaborator

BrianSong commented Jul 29, 2022

Hi @jiyongjung0, I wonder is this something we could support? I reached out to @venkat2469 and he redirected this to you.

@jiyongjung0
Copy link

@BrianSong This sounds like a good suggestion. Although I don't have much knowledge about conda, let me discuss with release engineers about how much effort will be needed to maintain the release.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants