Re: [Gnash-dev] AGG renderer complete

[strk]

Good work Udo. I suggest you apply the new code to the 0.7.2 branch
as well, it will be a magnet for bug reports ;)

--strk;

On Tue, Oct 31, 2006 at 10:09:35AM +0100, Udo Giacomozzi wrote:
> Hi all,
> 
>   my last commit to CVS (not 0.7.2 branch) implements all sorts of
>   masks for the AGG renderer, including nested ones. With this one
>   last change the renderer should support *everything* correctly,
>   except for the yet missing video support.
> 
>   There is a known issue with adjacent shapes not being perfectly
>   rendered (only for the anti-aliasing pixels), but it's only visible
>   in some rare cases.
> 
>   However, all gradients, gradient overflows, transformations, curves,
>   sub-shapes, masks, glyphs, outlines, cursors..... should be rendered
>   correctly. In the case of masks it may support even more than the
>   Flash plugin does (for example, text boxed are masked, while Flash
>   simply does not render them at all when they're under a mask and
>   AFAIK dynamic text does not work as a mask, while Gnash allows it),
>   however outlines have been explicitely disabled inside a mask, to
>   be compatible with Flash.
> 
>   The renderer uses damn much templates (note the AGG render handler
>   is my first c++ code I wrote ever) and this probably results in a
>   very bloated machine code. But it's a good thing because it speeds
>   up rendering (specializations for many situations).
> 
>   Anyway, the AGG renderer re-implements many parts of the rendering
>   process. First of all, it skips the tesselator. So, if you notice
>   any errors while rendering that work with other renderers, please
>   let me know.
> 
>   Otherwise I'll see to optimize the code a bit (possibilities noted
>   in the code) and have a look at the ActionScript classes because
>   there is certainly a lot left to do ;)
> 
> Udo