On Fri, Aug 27, 2021 at 10:26:39PM +0200, Paul Gevers wrote: > Hi, > > Sorry my previous message was weird. > > On 27-08-2021 22:11, Paul Gevers wrote: > > On 27-08-2021 21:58, Antonio Terceiro wrote: > >> One thing that happens when you do this type of change without > >> coordination is that all CI pipelines on unstable where rabbitmq-server > >> is installed are now broken. For example all merge requests against > >> debci at the moment have their tests in "failed" status. This creates > >> unnecessary noise for a lot of people. > > > > rabbitmq-server already got an update, so unstable should be fine (if > > not, shout (or better, file bugs)). I expect you mean testing, as I > > think that the point is that erlang already migrated before the issue > > was detected, otherwise an RC bug would have prevented the migration. > > > > That's why it was suggested earlier that rabbitmq-server should grow an > > autopkgtest as that have would prevented the migration. > > What I should have said: > we could have prevented migration of erlang until the reverse > dependencies were ready by having an RC bug on erlang. That would have > been totally appropriate if it would have lasted an reasonable time. I > *think* rabbitmq-server has problems migrating now *because* erlang > migrated, but that should clear up once the references are tested again. > However, it *also* has issues with being uninstallable.
FWIW, I just did that: I made a new rabbitmq-server upload adding a superficial autopkgtest to rabbitmq-server that just checks if the service is running after installation. This should avoid testing being broken because erlang migrated before rabbitmq-server has been fixed.
signature.asc
Description: PGP signature
