Fwd: kdebase/kate/app

[Matt Rogers]

On Thursday 21 April 2005 07:48 am, Christoph Cullmann wrote:
> On Thursday 21 April 2005 01:37, Matt Rogers wrote:
> > > Actually, we need working prototypes now, the more is better.
> > > I spent three days on my stuff, looks like Christoph didn't spent his
> > > entire life either ;)
> >
> > well, kdock* and kmdi is already a working prototype. yes, there are
> > things that could be done better, but isn't that always the case with
> > prototypes?
> >
> > :)
>
> but their design is flawed, it will be never possible to get any flawless
> session restore with the dockwidget design, without massive hacks, my
> approach only needs one reparent at most on session restore (and this will
> be eliminated soon, too), and no widgets
> flicker around on screen in the wild, which is really not solvable with the
> kdockwidget design, perhaps qdockwindow based kdockwidgets would make
> better, but not sure if qdockwidget really gives us the power to create the
> idea style gui at all
>

hence a new design will be needed, which is why i'm hoping for a new MDI 
library for KDE 4. While I understand the current resistance to the reuse of 
KDock* and KMDI w/o a complete rewrite, I still see some potential there for 
it to be useful again. I think all it needs is some love. :)


> > > > I still think that KDockWidget and current KMDI and friends are
> > > > savable.
> > >
> > > My experience and also Christoph's experience shows that ui library is
> > > not that hard to implement properly. KDock* and KMDI was there for 5
> > > years and they never worked as wanted. Therefore I always tend to
> > > question the ability to save them ;)
> >
> > the problem is that they were left to bit-rot and not properly cared for.
> > both yours and cullmann's code will suffer from the same problem
> > eventually as well if they're not maintained.
>
> that's true, for sure, but even if I have all features in the code to make
> me happy for kate, which would mean atm only some menu entries for
> show/hide toolviews + some context menu for the buttons to move the
> toolviews to a different sidebar, we are still only around < 1000 lines of
> code, from which say 500 lines are only wrappers around q or k widgets,
> with some minor additions, nothing real hacky, atm only hack at all is the
> save/restore of the splitter sizes, but this is a 30 lines hack, no evil
> event filter magic at all in play atm, and don't think will need it.
>
> cu
> Christoph

You also have to take into account who wrote them originally and who continued 
to hack on them. I can take one look at KMDI at least and know straight away 
it was written by a Windows programmer (no offense to anyone here, btw) and 
normally that's fine, except for the hack that nobody took the time to 
refactor it, and care for it. I still think the code is salvagable, but only 
time will tell. In the mean time, I'm interested to see what both you and 
adymo (and whoever else) come up with. :)

Matt
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
Url : http://mail.kde.org/pipermail/kmdi-devel/attachments/20050421/10d8e085/attachment.pgp