We discussed watching the semi-automated releases and letting them bed in before automating the later stages. The remaining uncertainty is around the specific commands to run. This change keeps the existing behaviour but:
Watches the PR to see when it's approved
Watches the tests to see when they pass, providing the URL to any failed tests so they can be re-run
Once the PR is in a releasable state it outputs commands that can be run to complete the release - this allows us to test that those commands would correctly create the release if they were run automatically
This allows us to keep the final release as a manual step but to prove whether the automation would handle it correctly or not.
The big benefit of this change is that we can see and refine the commands before we decide to run them automatically.
We discussed watching the semi-automated releases and letting them bed in before automating the later stages. The remaining uncertainty is around the specific commands to run. This change keeps the existing behaviour but:
This allows us to keep the final release as a manual step but to prove whether the automation would handle it correctly or not.
The big benefit of this change is that we can see and refine the commands before we decide to run them automatically.