Re: [Gnash-dev] open source AVM2 VMs

[strk]

On Wed, Sep 01, 2010 at 08:29:15PM -0600, Rob Savoye wrote:
>   So I've spent a little time looking at AVM2 VMs that are also open
> source. It seems that what we want to do is pass off any swf bytecodes
> Gnash sees to another VM, and get back something like a display list. As
> others have looked into this in much more depth than I have, I can't
> barely wait to be corrected. :-)

I belive the task of a VM is just to interpret bytecode and invoke
exterior stuff for any method call or variable get/set.
Should not have any concept of a "display list" at all.

> One thing to note is both ActionScript and JavaScript are
> based on the same ECMA standard, so a VM can supposedly be used for
> either. The biggest difference of course is that JavaScript is headless,
> and does no graphics, and flash of course, does graphics.

You can't compare JavaScript with Flash, but with ActionScript.
Both are headless/no graphics. But you can do graphics with both
by calling methods exposed by the hosting application (usually
a browser for JS and a flash player for AS).

Great work with the VM implementation research !

> SpiderMonkey
> ------------
>   SpiderMonkey is the current JavaScript VM in Firefox, and has the
> advantage of still being actively maintained. One interesting thing is
> Adobe uses SpiderMonkey as it's server side ActionScript VM for their
> Adobe Media Server. http://www.mozilla.org/js/spidermonkey/

Source for this Adobe/MS use of SpiderMonkey ?
All I find is vague mention of usage by Adobe [1] and wikipedia
mentioning it used in Acrobat [2] and nothing Adobe-else.

[1] http://partners.adobe.com/public/developer/opensource/index.html
[2] http://en.wikipedia.org/wiki/SpiderMonkey_%28JavaScript_engine%29#Usage

>   I guess the only other choice would be to use the AVM2 specs, now that
> they are available, and write our own based on that. Considering I tried
> funding two attempts at that which both failed and have had their code
> removed, make me not really sure that's a good option.

I'm not sure the problem was with the bytecode interpreter itself
rather than with the player API exposed to it.
But bwy will know better. I'd love to hear from him.

--strk; 

  ()   Free GIS & Flash consultant/developer
  /\   http://strk.keybit.net/services.html