search

h2oai / datatable

A Python package for manipulating 2-dimensional tabular data structures
https://datatable.readthedocs.io
Mozilla Public License 2.0
1.82k stars 157 forks source link

Pypi pipeline #3397

Closed GentleGhostCoder closed 1 year ago

GentleGhostCoder commented 1 year ago

It would be great if new versions were automatically pushed into the pypi repository. Currently, I get installation errors (from pypi) and therefore always have to install from github. Here is an example how to do github-actions to push releases to pypi (macos/windows/manylinux).

oleksiyskononenko commented 1 year ago

Thanks for your PR. We are currently working towards the new datatable version 1.1.0, but before pushing release to pypi we have to complete all the tasks needed for that version: https://github.com/h2oai/datatable/issues/3046

Meanwhile, with every datatable fix/enhancement/new feature/etc. we're building dev wheels for macOS/linux/Windows and publish them in our S3 repo and AppVeyor, see https://datatable.readthedocs.io/en/latest/start/install.html#install-latest-dev-version for more details.

1) In this context, could you please clarify your suggestion? 2) Also, what are the errors you are getting when installing datatable from pypi?

samukweku commented 1 year ago

@semmjon is there an existing issue for this?

GentleGhostCoder commented 1 year ago

There aren't any direct issues for this, though some are likely related to it. e.g. this https://github.com/h2oai/datatable/issues/3268

GentleGhostCoder commented 1 year ago

The merge request is not quite enough. You still have to add the pypi token to the Project Secrets and possibly a Github token (if the release-draft action is to be used).

GentleGhostCoder commented 1 year ago

@oleksiyskononenko I understand your concern. Possibly a pull request was the wrong approach here. The error I get when installing has to do with "gendoc" and is no longer a problem in the current master. I just wanted to give an example of a pipeline instead. :) I've been waiting for a more up-to-date version in pypi for a relatively long time, which is why one possibility would be to work with smaller releases (continues integration).

oleksiyskononenko commented 1 year ago

@semmjon Yes, I see what you mean. Can you just install datatable from our S3 repo? The wheels we are producing now are the dev wheels, so they shouldn’t be directly published to PyPi for the moment.

GentleGhostCoder commented 1 year ago

Yes, no problem. I just created my own Pypi project, where I can already push previous releases. I'll use my own project (repo) first. If you do the next release I might switch again. The pull request can be closed :)