Why is KConfig tier2?

[David Faure]

Le mardi 2 juillet 2013 21:19:29 Alexander Neundorf a écrit :
> On Saturday 22 June 2013, Aleix Pol wrote:
> > On Sat, Jun 22, 2013 at 1:17 PM, David Faure <faure at kde.org> wrote:
> > > Le samedi 22 juin 2013 13:03:22 Aleix Pol a écrit :
> > > > Hi,
> > > > We've been looking through some code scratching our heads on how we
> > > > can
> > > 
> > > do
> > > 
> > > > it to workaround uses of KConfig because of it being tier2.
> > > > 
> > > > Then we decided to look at it and we saw that KConfigCore and
> > > > KConfigGui only seems to depend on Qt.
> > > > 
> > > > So, maybe it could be moved to tier1?
> > > 
> > > Yes, it's tier1 now, since I put QLockFile in Qt.
> > > 
> > > (It was using kcoreaddons before, for KLockFile).
> > > 
> > > Back then I thought "let's not move it, all this will move again
> > > anyway",
> > > but
> > > clearly it's confusing people, so yes, please move it :-)
> > > 
> > > --
> > > David Faure, faure at kde.org, http://www.davidfaure.fr
> > > Working on KDE, in particular KDE Frameworks 5
> > 
> > Ok, so yes, it's misleading to have it in tier2. Or well, are tiers
> > documented anywhere? We've been looking for it too and haven't found
> > anything yet.
> 
> kconfig is still in tier2/.
> As far as I see, it depends on:
> - Qt
> - e-c-m
> - kdeqt5staging
> - KDEWin

The last two are temporary, so yes, it's tier1.

> So, should it be moved ?
> Is this simply
> git mv tier2/kconfig tier1/ ?

Yes (and adjusting any place that might still say "tier1/kconfig").

> Beside that, is anybody using Windows working on frameworks ? IIRC the plan
> for KDEWin was to get rid of it, right ? We also still need the XP manifest
> stuff for add_executable() AFAIK.

Getting rid of kdewin: yes. This is why David Gil Oliva is porting code away 
from time_t, among other things.

-- 
David Faure, faure at kde.org, http://www.davidfaure.fr
Working on KDE, in particular KDE Frameworks 5