D-BUS required changes in KDE 4
faure at kde.org
Fri May 26 17:02:20 BST 2006
On Friday 26 May 2006 17:45, İsmail Dönmez wrote:
> Thiago Macieira wrote On 26-05-2006 18:37:
> > David Faure wrote:
> >> On Friday 26 May 2006 11:35, Thiago Macieira wrote:
> >>> David Faure wrote:
> >>>> On Friday 26 May 2006 11:20, Thiago Macieira wrote:
> >>>>> There are two things I have found out during my port of kdelibs to
> >>>>> D-BUS that we will have to address:
> >>>>> 1) Currently, the D-BUS session daemon isn't automatically started.
> >>>>> This requires the user to have his scripts set up to start the
> >>>>> daemon *before* any KDE application is started. For a KDE session,
> >>>>> we can fix that with startkde, but imagine a user running KMail in
> >>>>> TWM: KMail will refuse to start.
> >>>> We currently have code to start kdeinit (which starts dcopserver)
> >>>> when a kde application doesn't manage to contact dcopserver (cf
> >>>> KApplication::startKdeinit and KApplication::dcopFailure). Can't we
> >>>> do exactly the same with DBUS?
> >>> We could, but why not do it in the libdbus-1 library itself and let
> >>> all programs using D-BUS to benefit from it?
> >> Sounds good.
> > That does answer the question, though: is it a feature we should push for
> > in D-BUS or should we simply mandate that the daemon be running in all
> > circumstances?
> The latter, distros should take care of that in init scripts.
Which distro are we talking about, when thinking about Solaris, Mac OS X and Windows,
for example? This isn't about Linux only; we need a dbus server everywhere.
Isn't the best way to ensure that we get one, to start one when needed?
Of course on linux we can expect that distros will launch a dbus server before kde,
but a fallback solution is very much necessary for all the other cases.
David Faure, faure at kde.org, sponsored by Trolltech to work on KDE,
Konqueror (http://www.konqueror.org), and KOffice (http://www.koffice.org).
More information about the kde-core-devel