Open Kalmalyzer opened 4 years ago
... the reason we need this test in the first place, is that we trigger a release by pushing a tag, but we fetch the build configuration (including version number) from the repository's contents.
If we had a different mechanism - not pushing a tag, just signalling "it is time to kick off a release" - perhaps a repository_dispatch event after all - then perhaps we could have the tagging as part of the release process, handled by the workers. They could tag when a release has completed successfully for example.
Then again, repository dispatch events require a token. Managing tokens is messy. make release
, perhaps that command should not take any input args?
If the release tag mismatches the Config.mk setting, strange errors can occur. We should have a test at the beginning of publishing that ensures these two are in sync.