This pull request addresses #27 by implementing a publishing pipeline in .github/workflow/publish.yaml.
The action is triggered every time a new version of lie_learn gets published and it automatically updates the entry on pypi.
In this way, downstream libraries (like escnn) can use lie_learn much more easily.
To fully close #27 it is required to:
1) enable trusted publishing
2) either publish a new release of lie_learn or manually trigger publish.yaml via:
Actions->'Build and Publish'->Run Workflow
Due to the end of life the current process only supports python>=3.9, but it is possible to include older versions to this workflow if needed.
Finally, notice that there are some issues with the recent release of Numpy 2.0.0 (also discussed in #30), so we have enforced numpy<2 in the dependency.
This pull request addresses #27 by implementing a publishing pipeline in
.github/workflow/publish.yaml
. The action is triggered every time a new version oflie_learn
gets published and it automatically updates the entry on pypi. In this way, downstream libraries (like escnn) can use lie_learn much more easily.To fully close #27 it is required to: 1) enable trusted publishing 2) either publish a new release of
lie_learn
or manually triggerpublish.yaml
via: Actions->'Build and Publish'->Run WorkflowDue to the end of life the current process only supports python>=3.9, but it is possible to include older versions to this workflow if needed.
Finally, notice that there are some issues with the recent release of Numpy 2.0.0 (also discussed in #30), so we have enforced
numpy<2
in the dependency.