[PATCH] Support for KUrl in KConfigXT and KCModule's

Andreas Pakulat apaku at gmx.de
Tue Mar 20 00:24:44 GMT 2007


On 19.03.07 12:18:25, Aaron J. Seigo wrote:
> On March 19, 2007, Andreas Pakulat wrote:
> > On 19.03.07 11:33:17, Aaron J. Seigo wrote:
> > > On March 19, 2007, Andreas Pakulat wrote:
> > > except for the added overhead of the call since KUrl needs to be fast. it
> > > doesn't look like the overhead is all that great, but still =) though,
> > > yes, i do think we should register these things just to make application
> > > code easier to write. i don't like the idea of such magical lines needing
> > > to be uttered everywhere.
> >
> > I'd like to note that kurl.h already uses the Q_DECLARE_METATYPE macro,
> 
> i know; i put it there ;) i originally put the matching qRegisterMimetype() 
> method in the .cpp but that was deemed unecessary by ... um ... i forget who 
> exactly. i would like to see this (and similarly important KDE types) 
> automatically registered, however.

Well, according to documentation qRegisterMetatype is only needed when
using queued signals/slots, but I think the Qt docs are not accurate
there as Qt complains when using KUrl with QVariant without the call...

> > > one place that might be of lowermpact than kurl's ctor is in
> > > KApplication? we could register the qvariants we want to be considered
> > > 'first class' items in kde apps. kurl would be amongst those, imho.
> >
> > Sounds like a plan.
> 
> great =) you or me?

I'll do it right away, including KUrl::List.

Andreas

-- 
You should go home.




More information about the kde-core-devel mailing list