search

graphql-python / graphene-tornado

MIT License
52 stars 14 forks source link

Publish graphene-tornado to Conda Forge #35

Open kinow opened 5 years ago

kinow commented 5 years ago

Hi,

We used this recipe for graphene-tornado, derived from the PYPI package, in order to publish our package to a private channel in Anaconda Cloud (we needed every dependency available in Conda Forge or channels).

The next dependency in our software that is missing a conda package is graphene-tornado. I will open the issue in Conda Forge near the end of this week. Is there any interest of one of the maintainers here in being added as maintainer in the Conda Forge repository?

graphene, graphql-core, and graphql-relay are already there.

Thanks Bruno

ewhauser commented 5 years ago

Can this be added to Travis CI so that the packages get published automatically? I've been meaning to do this for the PyPI packages. If so, would be happy to merge a pull request.

kinow commented 5 years ago

That's a good question. I only know the less automated way. Where someone publishes to PYPI, then the conda bots notify the conda recipe maintainers, that then update the recipe.

I am not sure if that's doable, as the conda recipe has no guarantee to work with the updated PYPI project release. e.g. if you change anything in install_requires for you setup.py, the meta.yaml of the conda recipe would also need to be updated to reflect this change.

I will take a look to see if there is any way of doing this too, as we will have to maintain quite a few packages in both PYPI and Conda Forge... so having more automation on this would help us too.

kinow commented 5 years ago

Submitted first version to conda forge. The Travis automation can be implemented afterwards (we will be maintaining ~4 packages in PYPI and Conda Forge, so we will also try to automate as much as possible). Let me know if I should add anybody else as maintainer.

Maintainers are notified for issues like pending releases to Conda to match PYPI version, issues, pull requests, and are able to release to Conda Forge.