Re: [Gnash-commit] OpenVG postponed for a week

[Rob Savoye]

On 08/14/11 14:17, Sandro Santilli wrote:

> Best if you can give a summary of what changes API-wise, which is
> not straightforward to get with all those master-merges.

  It's looking like a big enough task that I'm debating doing it post
merge. As to how Styles works now for OpenVG, only OpenVG/Framebuffer
users would notice the difference, and that would only be if they read
the code, as the current OpenVG code does work. So I'm not so sure it's
necessary for the merge to be committed. I'm gonna wait till I get back
in a week anyway. :-)

  I've been running an installed version of the merge for a week with no
problems. That and testing on a bunch of machines.

> I noticed a new getAveragePixel() in Renderer class for example, moved
> from MovieTester. Is that related to OpenVG ?

  No.

> Also, I've noticed a new Renderer::addRenderImage(GnashVaapiImageProxy*)
> which seems to be Vaapi-specific code for what should be a generic class.

  Could be, future plans are to refactor the vaapi code to be more
portable and useful. That's a future task though...

> I'd love to know more about your findings here.  What was AGG specific ?
> Parsed styles or rendering interface for it ?

  If you look at Renderer_agg_style.h, you'll see a bunch of templates
and other stuff that goes along with it. Some of that code has agg
specific function calls in it, but much looks generic. Our OpenGL and
Cairo renderers don't use the "new AGG Style" either, so I don't think
it's a big deal.

        - rob -