Hi, > as you seem to ignore mails by RT members, let me repeat my concerns > once again.
I am not ignoring any emails. It's just that many replies jump to discussing individual Salsa CI jobs or individual packages using them and not about the purpose of a CI or what it means to pass or fail. ... > > Specifically I'd like to float the idea that IF a project uses Salsa > > CI, AND IF the Salsa CI was failing (red) for the git tag > > corresponding to the upload (vcswatcher already has this data), the > > migration would be delayed by 10 days. > > Britney is concerned with what is in the archive. Salsa is not the > archive. It also has a questionable benefit if the Salsa CI per default > runs autopkgtests (which we already have covered), runs piuparts (which > we already have covered), runs blhc (which is unmaintained). Personally all my packages are green on Salsa CI before I upload them, and my error rate on the actual ci.debian.net is very low. I just referenced an example where an autopkgtest on Salsa CI that then occurred on ci.debian.net and is now blocking other packages instead of remaining an isolated case for the individual package that needs more work before being uploaded. You are now just dismissing that the autopkgtest in Salsa CI is useless. If you think Salsa CI is useless, why don't you then suggest people stop using it? If you read my suggestion and think about it a bit more instead of just jumping to dismiss it and discussing individual jobs, you hopefully realize that if implemented it would make some maintainers/packages disable Salsa CI, which would improve the situation for everyone as there would be less useless/wasteful Salsa CI use going on.
