Hi! On Mon, 2 Nov 2020 at 14:21, Dmitry Shachnev <mity...@debian.org> wrote: > > Hi Paul! > > On Sun, Nov 01, 2020 at 09:30:19PM +0100, Paul Gevers wrote: > > The trigger normally doesn't change which packages get installed, just > > their version. It's possible that during a transition, apt-get finds the > > wrong solution for the desired set, but once everything is rebuilt, that > > problem normally goes away. So, if the test is run against > > qtbase-opensource-src-gles, than either that's required due to a Depends > > or a test dependency. > > > > britney also adds triggers for source packages involved in versioned > > Breaks/Replaces and the like if those are different for testing and > > unstable. > > > > I hope this clarifies? > > This makes sense, thanks! > > Lisandro found the reason — that test is failing because it depends on > qt5-default package, which we removed. He filed #973607 for that issue.
In fact I did not mention this because I managed to fix the packages reported by dak in time for the transition. Yes, I am an old schooler that was used to use dak for this cases... Now I've learned that this is no longer true, as dak does not takes into accounts autopkgtests (but maybe it should? Or maybe -Rnb is not enough?). > It was not easy to find this reason because the test log doesn't even > mention qt5-default. This is indeed strange, as the package is clearly listed in tests/control. -- Lisandro Damián Nicanor Pérez Meyer http://perezmeyer.com.ar/ http://perezmeyer.blogspot.com/
