On Wed, Oct 20, 2010 at 5:14 PM, Rob Savoye
<address@hidden> wrote:
On 10/20/10 09:01, Alessandro Pignotti wrote:
> You could argument that I should have worked on Gnash in the beginning, but I
> wanted to build something new: a technology platform that could really be
> future proof. At this point I can say to have partly succeeded as I'm very
> happy of the graphics handling (especially the new engine not yet merged in
> master) and threading approach, while I'm not completely satisfied by the LLVM
> backend, and I'm considering switching to Chromium's V8 (if a suitable AST
> could be generated from the AVM2 bytecode).
I've considered V8 as well, but as you said, it would have been nice
if this had been added to Gnash instead of starting an entire different
project. :-)
> I've not any money to offer, but I'd like to invite all the devs to take a look
> at lightspark code and join our efforts.
Based on prior discussion, I see two problems with using LightSpark,
although that would be the preferable solution. One is that only
supports OpenGL, and the other is it uses LLVM. This leaves LightSpark
as a mainly desktop application. Me, I think mobile and embedded is the
way to go, so we need an AVM2 VM that supports smaller devices with less
footprint and doesn't need OpenGL.
The other problem is handling multiple versions of flash, which is
common. LightSpark only handles avm2, and hands off any older AVM1
content to Gnash. This would need to be changed to allow Gnash to hand
off AVM2 content to LightSpark. In this example though, I'd want a
display list of some kind that could then be passed back to Gnash for
rendering and other things.
- rob -