After several discussions trying to figure it out,
here is what i understand for now:
please correct me if i'm wrong.
* If you want to target new VM capabilities you'll have to use AS4 and
use stage3D for rendering (meaning using starling translated to AS4, to
render your display list)
* If you want to use current flex SDK on VMNext, you'll have to
translate it to AS4, and there is no easy way to do so, because of too
much differencies. Even using falcon to compile AS3 to AS4 seems to be a
big task which would compete with a new framework re write from scratch,
in terms of time effort.
* VMNext is architectured for games and theres a risk that it lacks some
flex capabilities like accessiblity.
Flex could have benefited of new VM capabilities if the new VM would
have been an enhancement of the old VM, but it is a complete replacement
with new language so its not the case.
Thats why we can't expect any runtime enhancements for flex if we don't
target new runtime.
* The next versions of flash player and AIR will embed the new VM but
also the old ones, meaning that any new device (like windows 8)
supported by new AIR runtime will also support flex apps. I would be
glad to get a clear reassurance about this fact, but according to
Adobe's roadmap and other annoucements, it seems to be the case for at
least next 5 years.
I'm still waiting an answer about the captive runtime thing for VMNext.
* Thats why IMO, targetting VMNext is not quite a priority for flex. And
it does not worse the effort.
* It get us back to the main discussion, which is how to make flex cross
platform again and without the strong dependency with Adobe's runtimes.
It seems that until today, some commiters like Alex consider the actual
code base too monolithic (it still brings some old code made to optimize
for AS2) to be able to refactor it as we need, even breaking UIComponent
seems to be quite difficult.
And thats why a complete rewrite with cross platform in mind would be
the wisest choice.
We still have to figure out how to do it, especially we have to figure
out if it must be written in AS3, or Haxe or AS4, or anything else?
* For the current flex SDK based on Adobe's runtimes, it still can be a
good framework to manage the transition.
* Making current flex render on starling is still a valid short term
goal because it would give better performances on current and next
devices, especially for retina display if i'm not wrong. And it seems
feasable but it does not suites to a minor update of the framework.
Le 15/11/2012 11:04, Sebastian Mohr a écrit :
Hi there,
Does anyone have any insides of the new AVM of the forthcoming
Flashplayer 12 in 2013? Are there any APIs (e.g. specification documents)
that Adobe is willing to share? Is Starling [1] the only hook we can get?
Sincerely Yours,
Sebastian Mohr
Apache Flex Developer (PPMC),
Interaction Designer & Musician
http://www.linkedin.com/in/masuland
[1] http://gamua.com/starling/
