Feature Type

  • [X] Adding new functionality to pandas Github Actions

Problem Description

The many github actions associated with pandas (seem to) individually declare variables that are conceptually common across workflows e.g., minimum supported python version.

This is a maintenance headache which can be resolved by storing our environment variables in a file and utilising a GitHub Action to fetch those vars for GitHub workflows.

Feature Description

Here is a good outline of how to implement a shared, reusable item across github workflows.

https://arinco.com.au/blog/github-actions-share-environment-variables-across-workflows/

Alternative Solutions

  • Ignore best practice as too much effort. Sounds snarky, but I do mean it as a viable solution. This can remain a future maintainer's problem.

Additional Context

No response

Comment From: lithomas1

Hmmm, I'm not sure how I feel about this one.

How about we push some of those variables(e.g. COVERAGE, PYTEST_TARGET, TARGET?) down into ci/run.tests.sh? We can set default values for those variables if unset, and then override if necessary in the Github Actions config.

Comment From: mroeschke

Yeah I would be -0.5 refactoring env variables into separate env files (I haven't seen other pydata project use this approach).

Another option would maybe create a github composite action to set env variables?

Comment From: mroeschke

Thanks for the suggestion but since this issue hasn't had too much core dev buy-in closing for now