[EMAIL PROTECTED] wrote:
> libart defines an ArtRender structure that is responsible to render objects
> on a buffer, like a context. So I attach an ArtRender to the ggi_visual, make
> it render directly in the visual framebuffer, and define wrappers around the
> art_render_* api functions that take a ggi_visual instead of an ArtRender.
>
> For now it only works if a linear framebuffer is available for the visual,
> like for GGIMesa.
>
> The only thing is that the rendering process is quite slow (cos libart is slow)
> and I'm not sure if it can be easily accelerated. I just defined default stubs
> that call libart_lgpl functions.
>
> I still have a little thing to fix and I'll make a first version available.
what you describe is more or less what we do in the libArt based DrawingKit we
use in the berlin project (beside the MesaGGI based one). Because of some subtle
'incompatibilities' we have to draw into a memory-visual and once the screen
repair is over, we crossblit from that to the screen visual.
I think there are a couple of hot spots which, if properly worked out, could
make libart-on-GGI pretty much faster.
Regards, Stefan
_______________________________________________________
Stefan Seefeld
Departement de Physique
Universite de Montreal
email: [EMAIL PROTECTED]
_______________________________________________________
...ich hab' noch einen Koffer in Berlin...