seamless project model update

[Aleix Pol]

On Sat, Aug 7, 2010 at 4:15 PM, Andreas Pakulat <apaku at gmx.de> wrote:

> On 07.08.10 14:00:54, Milian Wolff wrote:
> > On Saturday 07 August 2010 13:45:00 Andreas Pakulat wrote:
> > > On 07.08.10 13:30:12, Milian Wolff wrote:
> > > > On Monday 12 July 2010 13:08:25 Aleix Pol wrote:
> > > > > It was 4am and Q_DECLARE_METATYPE was giving me problems. yes it
> should
> > > > > be possible
> > > >
> > > > *revive*
> > > >
> > > > what happened to this? I'd still like to see the seamless update very
> > > > much :)
> > >
> > > I posted the results of my experiments in another thread afaik, its not
> > > useful to add locking to the project model (and also not really doable
> > > as we don't control the complete API).
> > >
> > > And 'scheduling' the addition to the model into the main thread would
> > > require the foreground lock from the movingranges branch (I don't see
> > > any necessity to replicate that code just for this single feature).
> >
> > OK, thanks. Then let's wait for movingranges branch to be merged.
>
> Actually I mis-remembered what I did myself :) If you look at the code in
> the model you'll notice that adding items from a background thread is
> possible already (invokeMethod is being used) and I think there's even a
> test for that (though I don't know if it fails without the invokeMethod).
>
> Andreas
>
> --
> Your boss climbed the corporate ladder, wrong by wrong.
>
> --
> KDevelop-devel mailing list
> KDevelop-devel at kdevelop.org
> https://barney.cs.uni-potsdam.de/mailman/listinfo/kdevelop-devel
>

If you add an element without invokeMethod from a thread that's not the same
as the view the signals are emitted in the wrong order (well, correct order
but with the rowCount information in the model already up-to-date which
creates the annoying empty cells).
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/kdevelop-devel/attachments/20100808/1bc230a7/attachment.html>