Control: forcemerge 748668 768774 Control: retitle 748668 slim: Under systemd, randomly hijacks default-x-display-manager ignoring default selection Control: affects 748668 gdm3
On Sun, Nov 09, 2014 at 06:40:40PM +0100, dAgeCKo wrote: > Le 09/11/2014 17:47, Michael Biebl a écrit : > >Control: reassign -1 slim > > > >Am 09.11.2014 um 09:37 schrieb dAgeCKo: > >>Package: systemd > >>Version: 208-8 > >>Severity: normal > >>Debian: Testing amd64 > >>Regression: No > >> > >>If several display managers are installed (for example gdm3 and slim), > >>at each boot, systemd seems to randomly choose one and launch it. > >>So we might have gdm3 at one boot and slim at another boot. > > > >That's not a bug in systemd, but slim > >In Debian there is traditionally a /etc/X11/default-display-manager > >config file, which describes the default display manager. > > > >The sysv init script for the display manager reads that file, checks if > >it's supposed to start and exit's otherwise. This means, you can have > >multiple display managers installed and all of their sysv init scripts > >enabled. > > > >slim does ship a native .service file, which doesn't include that check. > >It should add a check like gdm.service (or lightdm.service) > > > >ExecStartPre=/bin/sh -c '[ "$(cat /etc/X11/default-display-manager > >2>/dev/null)" = "/usr/sbin/gdm3" ]' > > > >slim should also setup the /etc/systemd/system/display-manager.service > >in its maintainer scripts. > > > >I'm re-assigning this bug to slim. Ideally it uses the same scheme as > >was implemented for gdm3 and lightdm. > > > >Joss and Martin have designed and implemented that scheme, so I've CCed > >them in case the slim maintainer has more questions. > > Thanks. Since slim and gdm3 behaved well when systemd was not used, I > quickly thought that systemd was the guilty... This has already been reported as #748668, thus forcemerging lightdm implementation is suggested there as the way to go for #748668, as well as the info under titanpad. That info should be enough for a fix, so I tagged that bug report as +patch even if no explicit patch was added. Since you experienced this problem also with gdm3 I am marking this bug report as affecting it. Should this be RC? Regards, -- Agustin -- To UNSUBSCRIBE, email to [email protected] with a subject of "unsubscribe". Trouble? Contact [email protected]
