Description
Description:
Given two workflows in the same repo using this action, have one workflow ignore the state saved by another. I think either of these two options would work for us:
- Make the cache key specific to the workflow
- Make statefulness optional with a config setting
Justification:
Over in Matplotlib, we are using the action to slowly work through a large backlog of issues. We want to keep notifications fairly minimal so have operations-per-run set low. However, we don't want steps like removing the label after update or closing the issue to wait until we've worked through the whole backlog. So when we updated to v9, we also added a second workflow using the action with days-before-stale=-1 and a much higher operations-per-run (see my PR here matplotlib/matplotlib#27523). The two workflows run on alternate days. Once #1136 is fixed, I believe what will happen is
- The original workflow processes a few issues and saves the state
- The second workflow finds that state, processes all the remaining issues and deletes the state
- The original workflow starts from the beginning
Since only the original workflow is adding the label, we still do not make progress through our backlog! If each workflow has its own state, this would not be a problem. Also if we could tell the second workflow not to look for a state, it would not be a problem.
Are you willing to submit a PR?
Willing, yes, but I do not have skills with typescript....
Activity