policy change related to kdelibs snapshots

David Faure faure at kde.org
Mon Jul 10 16:34:27 BST 2006


On Monday 10 July 2006 17:23, Benjamin Reed wrote:
> On 7/10/06, David Faure <faure at kde.org> wrote:
> 
> > I meant kde apps in kde subversion, but ok :)
> 
> Right, I was trying to show how our methodology would work in the
> context of kde development...
> 
> > Well, the whole idea of snapshot is to avoid intermediate changes, as they
> > would happen if tracking trunk all the time. More than once we changed our
> > mind -before- updating snapshot, and this didn't affect the apps *because*
> > snapshot exists. Not all changes can be done in a work branch, when they
> > affects a few apps with many dependencies.
> 
> My impression is that it didn't really end up working that way very
> much.  Perhaps I'm mistaken.

No, you're right, and that's what coolo's suggestion attempts to fix :)

> I think ultimately, the problem is coming from the "kdelibs-only"
> approach.  The apps reallypretty much need to be branched along with
> libs if API changes are happening, it's really the only way to make
> sure things stay in step.

And that's what bleedingedge/ is all about. Porting the apps at the same
time as changing kdelibs trunk.

> Putting everything in bleeding should in theory solve that issue, but
> it still seems like we're kind of ending up with 2 bleedingedge's, one
> more stable than the other, rather than unstable + something app
> developers can really use.  Granted, the dbus move was a special case,
> it was ripping out something that nearly every KDE app ended up using
> in one way or another.  No getting around it, things have to be broken
> for a while in that case.

Yes, but right now we're seeing other such changes. Widgets being removed and
then re-added when we figure out they are used after all :),
APIs being changed (no more K3WidgetAction - and there's a bigger action change
in the pipe if/when liveui delivers), etc.

> There's no 4.0 final API, but it sure seems like 2 weeks is pretty
> out-of-date at the current rate of development...  <grin>

It's recent enough for app developers to concentrate on app stuff
(akonadi, koffice flake, plasma, etc.) instead of spending all their time on porting
to kdelibs changes.

-- 
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 mailing list