KIO split (from the Job side)

Volker Krause volker.krause at rwth-aachen.de
Tue Apr 18 13:32:49 BST 2006


On Tuesday 18 April 2006 09:37, Kevin Ottens wrote:
> Le Lundi 17 Avril 2006 22:56, Volker Krause a écrit :
> > Looks good from the Akonadi point of view :)
>
> Great!
>
> > One thing we have in the Akonadi job class is a synchronous convenience
> > method:
> >
> > bool Job::exec()
> > {
> >   QEventLoop loop( this );
> >   connect( this, SIGNAL( done( PIM::Job* ) ), &loop, SLOT( quit() ) );
> >   start()
> >   loop.exec();
> >
> >   return ( d->mError == 0 );
> > }
> >
> > Might be interessting for KJob as well, it simplifies things eg. if you
> > are using the job in a thread.
>
> Yes, noticed this one. I'll add it as soon as possible.
> Just one question though, will it play nicely with deleteLater() though,
> since I use it in emitResult() (and KIO::Jobs are assumed to delete
> themselves).

Good point, I think it will work, but that's a question for a Qt event loop 
expert.

> > Hm, what about ThreadWeaver, IIRC it has a job class as well? Would be
> > cool if KJobs and ThreadWeaver jobs would be at least compatible, using
> > diffrent job systems inside each other is ugly, I know that from KNode
> > ;-)
>
> I discussed it a bit with Mirko, as far as I know it has mostly an impact
> on the scheduling of jobs, it's something not imposed by KJob. Currently I
> only want KJob to force a consistent job interface, scheduling and the way
> they are executed is a slightly different problem.
> I agree that at one point we'll surely want to address this, but it needs
> some refactoring first in order to use ThreadWeaver correctly (IMHO
> removing the UI specific parts of jobs is of importance here).

Sure, this was just a point that crossed my mind while thinking about the job 
stuff. 

regards
Volker
-------------- 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/kde-core-devel/attachments/20060418/68dcf025/attachment.sig>


More information about the kde-core-devel mailing list