Re: [Gnash-dev] one day AVM2? who knows

[Rob Savoye]

On 11/29/11 11:57, address@hidden wrote:

> Well, while it would be nice to use lightspark as a library, given
> that this won't happen, the question to ask is: how much effort would
> it be to *fork* the lightspark code, to use as a base for AVM2
> support in Gnash?

  I'm not sure LightSpark would be worth the effort. For one thing, it
uses LLVM and OpenGL only, which won't work on an embedded platform. I
can go into other technical issues why this wouldn't work well if anyone
wants. An supported a forked version of a project that is under heavy
development can be a maintainance nightmare.

> Tamarin is probably more mature (AIUI it's actual code from Adobe's
> proprietary player); but the fact that lightspark already implements
> the most important AS3 classes, might actually outweigh this. (If
> they can be integrated into Gnash in a sane manner...)

  Ideally we want a VM we can hand off byte codes to, and get back a
display list for Gnash to render. Tamarin is closer to working that way
than anything else I looked at. Tamarin has some AS3 classes, but not all.

> Or perhaps it might even be possible to do both: take Tamarin as the
> AVM2 engine, and the AS3 classes from lightspark?

  Possibly... Or from HAXE. I find it funny that the only Gnash task
that has funding already nobody wants to do... (at least to get it started)

        - rob -