(note to myself: you sound really grumpy after reading bikesheds) On Thu, Feb 13, 2014 at 04:31:36PM +0100, Andreas Beckmann wrote: > On 2014-02-13 15:37, Ondřej Surý wrote: > > On Thu, Feb 13, 2014, at 15:00, David Kalnischkies wrote: > >> On Thu, Feb 13, 2014 at 01:38:51PM +0100, Ondřej Surý wrote: > > >> Broken libdb5.3-dev:amd64 Conflicts on libdb5.1-dev [ amd64 ] < 5.1.29-7 > >>> ( libdevel ) > >> Considering libdb5.1-dev:amd64 -1 as a solution to libdb5.3-dev:amd64 > >> -1 > >> Holding Back libdb5.3-dev:amd64 rather than change libdb5.1-dev:amd64 > >> Try to Re-Instate (1) libdb-dev:amd64 > > >> If you care for an explanation of the output: > >> https://lists.debian.org/deity/2014/01/msg00133.html > > > > This really doesn't explain how to solve this problem and get the > > apt-get to upgrade libdb-dev. > > > So any ideas how to force the libdb-dev upgrade? I am going to remove > > libdb5.1*-dev in couple of days, but I just want to make sure that > > That might be sufficient to change scores to make the upgrade work > smoothly. Right now libdb5.1-dev gets a bonus point for being installed > and being installable/upgradable, but it won't get that any longer if it > is an obsolete package (no installation candidate available any more), > so libdb5.3-dev might win the "fight" instead of making a tie. > > I think the current scores of -1 are the result of > > 5.1: priority:extra = -2 > installed,installable = +1 > rdepends: 0 > ==> -1 > 5.3: priority:extra = -2 > installed,installable = 0 > rdepends: +1 > ==> -1 Indeed, that is the scoring with s/installable/downloadable/. And it is indeed fixing itself if 5.1 is no longer downloadable. The thing is that libdb-dev has basically no depends (it has some, but based on the popcon, most people must have it installed "standalone"), so it isn't in a good position to force other packages away aka everyone who has it "standalone" will have it autokept until the previous version is gone from the archieve. piuparts e.g. has the same problem, even through it isn't reporting it as an error as it seems (Andreas?). I am pretty sure we had talked about the libdb-dev depends libdb?.?-dev already a few times with the outcome that it is somehow needed to have multiple non-coinstallable libdb?.?-dev's around instead of doing the usual unversioned -dev package dance, so I considered it futile to do it once again and just commented on effect instead of cause. Or maybe I have mixed that up with boost and/or the other "offenders". Best regards David Kalnischkies
Attachment:
signature.asc
Description: Digital signature