[cairo] Re: [Mesa3d-dev] Points in Cairo Glitz paper

[David Reveman]

On Thu, 2004-04-22 at 07:18 -0700, Jon Smirl wrote: 
> --- Carl Worth <cworth at east.isi.edu> wrote:
> > As for cairo, if there is another way to draw the correct results in a
> > given backend, it's not necessary to use a low-level interface with the
> > same semantics as Render. (That said, the current OpenGL backend was
> > written with the same interface as Render and it seems to be performing
> > well.)
> 
> It's performing well on the highest end consumer graphics card in the market.
> Most 3D hardware out there has fewer texture units and no shader support. Glitz
> requires shader support. Glitz is also using OpenGL features that aren't
> supported in mesa yet.

Glitz performs quite well on older hardware as well, there are just some
things it can't do. e.g. offscreen rendering, anti-aliasing and
gradients. It can still very efficiently draw trapezoids and composite
surfaces. Whenever cairo wants to do something that glitz can't support,
software image buffers will be used, so it should never fail completely.

Offscreen rendering is important for cairo. Without offscreen rendering
cairo_surface_create_similar will not work and we can't do much else
than drawing trapezoids and composite surfaces efficiently. This is why
we haven't focused that much on supporting old hardware without pbuffer
support.

What we can do, is to add support for some gradients without using
shaders and add support for compositing with mask surface using only
multi-texturing. This shouldn't be impossible.

- David

-- 
David Reveman <c99drn at cs.umu.se>