Re: [Gnash-dev] Everybody talks about, but nobody is doing anything about it

[Udo Giacomozzi]

Hello Tomas,

Thursday, July 27, 2006, 7:53:03 PM, you wrote:
TG> So Gnash needs more backends,

Our company finally decided that we put time resources in the
development of a direct framebuffer backend (stand alone player only).
Hannes Mayr is on this list too and together we'll try to reach this
goal.

However, for any backend I still think it would be good to have a
short intro regarding the general structure of a backend from someone
of the maintainers.

A flash movie consists of many layers and movieclips. Is there some
counterpart in the backend? How are these cached, i.e. when one layer
changes, how is it avoided that another complex layer at the same
position (that did not change) is being redrawn? Or, simply put, how
is a frame generated (i.e. how's the backend being used)?
And so on...

TG> and the is a general agreement that a cairo based backed should
TG> be one of (if not) the first to be implemented, since it itself can use
TG> multiple backends. There is already a cairo backed, but it isn't really 
working
TG> (that's what I've heard anyway).

Cairo it's said (have no experience, though) to be rather slow.


TG> IMHO it would be better if we changed the structure so that we
TG> passed entire figures to the backends where possible (squares, cirkles, 
beizer
TG> curves etc.), and only use triangulation when absolutly nessesary. This 
would
TG> enable cairo to take advantage of it's backends, and other backend could do
TG> what they can, etc.

Agree, however I suggest some fallback mechanism.

Udo