Hi!
On this list we have had several discussions about CI. Historically there has been a debian/gitlab-ci.yml that ran a custom Go `test_the_archive` job on a custom Go runner. As discussed on this list several times in past year, it does not seem to provide much value right now, as it does not fail even if the package itself is unbuildable.
I think we have enough experience from the past year now to conclude that:
- Go team packages should run Salsa CI to actually build packages and get meaningful feedback
- Go team packages don't get additional value from the custom runner or custom `test_the_archive` job
- All packages should move to use debian/salsa-ci.yml as the CI/CD configuration path going forward
- ..and that file could be just a vanilla salsa-ci.yml file
To be able to move forward with this, I would need to have 3-4 people to Approve or at least thumbs up the PR at
https://github.com/Debian/dh-make-golang/pull/279
It feels to me that the Go team has no formal way of deciding on things. Having enough people "vote" on a PR I think would constitute enough of support to take a decision on something. If we can fix tools like `dh-golang` and `dh-make-golang` create new packages in a certain way that seems most optimal, that should be enough to signal what the team decision was. After that updating *existing package* en masse is just a smaller technicality of how to best execute it.