[Xfce4-commits] r25185 - in libfrap/trunk/libfrap/menu: . tests

[Brian J. Tarricone]

I totally agree with all the stuff Erik said.  And stuff.  While I
personally don't know whether or not I *need* 2.10, I have no problem
with depending on it if it will make the other devs' lives easier.

	-b

On Mon, 19 Mar 2007 21:46:26 -0400, Erik Harrison wrote:

> On 3/19/07, Jannis Pohlmann <jannis at xfce.org> wrote:
> > On Mon, 19 Mar 2007 23:15:59 +0100, Stephan Arts wrote:
> >
> > > On 3/19/07, Harold Aling <h.aling at home.nl> wrote:
> > > >
> > > >  Olivier Fourdan wrote:
> > > >  -----BEGIN PGP SIGNED MESSAGE-----
> > > > Hash: SHA1
> > > >
> > > >
> > > > Hi
> > > >
> > > > Stephan Arts wrote:
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >  Now that the most important menu spec tests are passed and
> > > > Brian already uses FrapMenu in xfdesktop, shouldn't we move
> > > > FrapMenu out of libfrap and into Xfce trunk? Of course this
> > > > raises the question about how we plan to reorganize our
> > > > libraries for Xfce 4.6 again - I'd be glad if we could come to
> > > > a conclusion here soon.
> > > >
> > > >  No opinions?
> > > >
> > > >  Well, if we can agree on which Gtk+ version will be the minimal
> > > > requirement for 4.6 (2.10 has been mentioned more then once on
> > > > this list) we can decide which stuff is deprecated regarding
> > > > this and perhaps some (new) freedesktop.org standards.
> > > >
> > > >  I'd really prefer 2.8 instead of 2.10, Edgy will ship with 2.8
> > > > and setting a dependency on gtk+-2.10 would exclude a large
> > > > part of our user base.
> > > >  Those on (x)Ubuntu Edgy probably won't install Xfce 4.6, as it
> > > > probably won't be available through the standard repositories.
> > > > Feisty Fawn (the soon-to-be current version) currently has
> > > > v2.10.11 of GTK+.
> > > >
> > > >  If Xfce 4.6 could really benefit from GTK+ 2.10, I'd say: go
> > > > for it!
> > >
> > > Afaik, olivier was refering to debian etch instead of ubuntu edgy.
> > > (especially the 'will ship' indicates he means a release which
> > > has not been made yet, edgy has)
> > > Debian currently has 2.10 in experimental, 2.8 is the version in
> > > testing and unstable.
> > >
> > > A feature we might want to have is 'recent document support'
> > > which is available since gtk 2.10. However, since 2.10 is not
> > > regarded stable (even according to gtk.org so it seems), I think
> > > olivier is right on the dependency on 2.8. We do not want a
> > > release which has all sorts of bugs simply because it is
> > > impossible to run on stable libraries.
> >
> > I'm fine with GTK+ 2.8 even though I would prefer to got for 2.10.
> > Recent documents is something we can live without for another
> > (hopefully short) release cycle, IMHO.
> 
> Honestly, we're talking about replacing MCS, moving to the long
> gestating LibFrap, we have no full time developers, and if Jaspers
> presence on this list is any indicator at least one of the major devs
> has less and less time for Xfce development. Hell, I'd be willing to
> guess we all have.
> 
> Mousepad, Terminal, and Thunar were all initially discussed as out of
> core development projects and have always jumped on using whatever Gtk
> and other libraries have offered, and will likely continue to, even
> though they are now considered "core".
> 
> Gtk has shortened it's release cycle since Xfce 4.2, and significant
> chunks of deprecated functionality come from our attempt to wrap or
> backport later Gtk functionality, which is tedious to maintain.
> 
> Xfce devs tend to run latest Gtk stable anyway, or close to it. I
> can't count the number of times one of us has accidentally introduced
> a dependency on a higher version of Gtk and not noticed - sometimes
> not till very late in the game since our users tend to be up to date
> as well.
> 
> Those distros which don't ship a Gtk 2.10 package likely won't ship an
> Xfce 4.6 package either, so our users on those platforms will be
> building from source anyway, and having to upgrade fundamental
> packages like DBUS and HAL as it stands while they're at it - why not
> Gtk+ the most fundamental Xfce dependency?
> 
> Due to the structure and manpower of Xfce we have no overarching goals
> for 4.6, rather a series of per project goals and Olivier as the final
> arbiter of what goes in and what goes out. That tends to slow the
> release cycle. Not only that we're starting this dev cycle the same
> time we started the last two, with the same questions and the same
> amount of manpower, more or less.
> 
> Xfce has a history of it's major releases being featureful and highly
> stable. Any shortening of the 4.5 development to less than a year is,
> in my opinion, just asking to suck away a lot of our time in the fall
> fixing bugs all at once that we usually manage to fix piecemeal over
> our nice peaceful unstable run, an additional burden on our small
> team.
> 
> In case you haven't noticed, this is a rather strong appeal to make
> Gtk+ 2.10.x the assumed minimum Gtk version for Xfce 4.6. Perhaps it's
> a bit fatalist, but we haven't seen the drastic shift in manpower or
> project vision that makes me reasonably think that 4.5/4.6 development
> will go any different than 4.3/4.4. And considering the quality of
> 4.4, I'm happy with that, but we should learn from history.
> 
> If we decide on 2.8, I'll wrap everything in ifdefs, but I honestly do
> not see a win for us at all. I'd be willing to be 20 American dollars
> that Gtk+ 2.16 will be out by the time Xfce 4.6 is, and that a
> majority of our users will be at 2.12.  We should use that to our
> advantage.
> 
> That's my wad of cash.
> 
> >
> >   - Jannis
> >
> > _______________________________________________
> > Xfce4-dev mailing list
> > Xfce4-dev at xfce.org
> > http://foo-projects.org/mailman/listinfo/xfce4-dev
> >
> >
> 
>