On Sun, 2017-11-26 at 17:53 +0100, Andreas Metzler wrote: > So essentially if mariadb-10.1 migrated a mariadb-10.1 with mariadb- > test would be replaced by a mariadb-10.1 without mariadb-test?
That appears to be the net effect, yes. > Given that afaict mariadb-test has no reverse dependencies could you > consider forcing the migration? > > I am asking since mariadb blocks exim, which urgently needs an > update. A test run suggests that hitting mariadb-10.1 with a large hammer has no negative effect beyond the -test binary package disappearing, so given the number of migrations that are currently blocked behind it (hurrah for symbols changes) I've added a force-hint for tonight's britney run, which should unblock things. (I'm still not entirely clear why britney isn't just doing the right thing here anyway.) Regards, Adam
