Re: Virtualizing paint functions

[Anders Carlsson <andersca imendio com>]

Owen Taylor skrev:
> On Mon, 2006-04-10 at 20:18 +0200, Anders Carlsson wrote:
> \
>   
> > > I'm basically not following what you are talking about at all, I think
> > > I missed a step earlier :-(
> > >
> > > Right now, you have a bunch of the functionality diverted to the
> > > GdkPaintable interface, but not process_all_updates(), right?
> > > So if you implement GdkPaintable on your window implementation, 
> > > process_all_updates() does nothing because nothing is added to
> > > the dirty list?
> > >   
> > >       
> > Yeah, that's right, and currently this is not a problem because the only 
> > place (in gtk+) where process_all_updates() is being called is in an 
> > idle function (scheduled by the standard invalidate_region() function), 
> > and for Quartz the invalidated regions will automatically be redrawn in 
> > the main loop.
> >     
>
> And at that point you generate expose events and call GTK+'s event 
> function?
>   
Yes.
> Does that also trigger if someone does
>
>  while (g_main_context_pending (NULL))
>    /* nothing */ ;
>
> as if trying to get things to update while doing something intensive?
>
>   
Yes, that should work just fine.
> > > What would be the problem of making process_all_updates() part of
> > > GdkPaintable? I don't think it is worth worrying about the case
> > > where a backend wants to implement everything else in GdkPaintable
> > > and not that ... we have *zero* examples of such a backend,
> > > right?
> > >   
> > >       
> > The problems is that gdk_window_process_all_updates doesn't take an 
> > argument, so there is no object passed in which can implement GdkPaintable.
> >     
>
> Ah. May not be worth fixing the problem until a need arises, but if 
> you wanted to do so, an optionally null function on GdkDisplay seems
> easiest ... then the current function could just walk over open
> displays.
>
>   
That sounds like a good plan for the future if necessary, yes. I'm going 
to commit the latest patch I attached unless there's anything wrong with it.

Thanks,
Anders