flex-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Gordon Smith <gosm...@adobe.com>
Subject RE: Flex 5 in haxe
Date Fri, 16 Nov 2012 19:27:45 GMT
> Since Falcon has little support of MXML, I see we don't loose almost nothing.

Given that Falcon can compile MXML apps like Checkinapp, I don't think it's accurate to characterize
its level of MXML support as "little". My own characterization would be "80% complete". But
completing the other 20% is a daunting task ( it's certainly many person-months of work) and
not enough people are stepping up to help. I'm guessing that this is because the major companies
who used to use Flex and swore that they had plenty of compiler engineers who would help finish
Falcon have moved on to other technologies.

So I agree that any and all compiler technologies, including Haxe, should get serious consideration.

However, I think before Apache Flex makes a compiler decision, it needs to decide what runtimes
it is targeting. Move to V12? Abandon Flash? Produce HTML/JS? Produce native iOS code? Produce
native Java for Android?

- Gordon

-----Original Message-----
From: carlos.rovira@gmail.com [mailto:carlos.rovira@gmail.com] On Behalf Of Carlos Rovira
Sent: Friday, November 16, 2012 2:49 AM
To: flex-dev@incubator.apache.org
Subject: Flex 5 in haxe


as we were discussing yesterday, there's room to a new Flex framework written from scratch.
As we don't want to rely in Adobe technologies anymore we were talking about haxe. We can
make it now that work would be starting from zero.

Haxe is a platform developed by Nicolas Canesse that made it's own community. Nicolas is a
genius of compilers. People coming from Flash Open Source will remember MTASC compiler back
in 2004-5. If you search and investigate you will found that haxe is very powerful and is
"the great unknown technology".


haxeNME is like Adobe AIR and seems to be more performant in iOS, and Android (see http://esdot.ca/site/2012/performance-showdown-starling-vs-nd2d-vs-genome2d-vs-haxe-nme).
Supports as well Windows, Mac, Linux and BB.


There's an haxe plugin for IntelliJ. But in my test it seems that only supports haxe and not
NME yet.

(Disclaimer: I'm to new to haxe and haxeNME and maybe I wrong making some statements here).

- Haxe is OOP and is "one language to rule them all" philoshopy.

- Haxe compiler is better that the set provided by Adobe (I'm referring to
AS3 legacy compiler. Falcon is new technology and maybe this is not true. I does not have
any info to make a comparision between falcon and haxe compiler).

- Haxe language is more evolved (maybe even Adobe AS4 will copy things from

- Haxe support HTML5/JS out of the box (but it seems to be in beta status).

- There's a Starling port in haxe.

Regarding Flex: haxe compiler could bring to flex things like *metadata
evolution* or *AOP*. Adobe compiler will never get that evolution since gamming is not focused
in that kind of things...This is more likely to see in haxe if Flex 5 works than expect it
from Adobe.


IDE: IntelliJ+haxe plugin. IntelliJ is the best option for Flex, and supports haxe, but I
think haxeNME is not supported yet. But IntelliJ guys are behind the plugin, so things could
evolve ok in this point.

MXML: I think there's nothing like MXML in haxe today, and this is one of the key points in
Flex. We would need to put the efforts of MXML in making it possible in haxe. We could talk
with Nicolas Canesse about this possibility. Since Falcon has little support of MXML, I see
we don't loose almost nothing.

So my proposal is:

* Start Flex 5 from scratch with haxe.

* Use the Flex 4 API to model Flex 5 over haxe (as the first draft).

* Start using Starling haxe library as the core displayobject API (to be able to target Stage3D/Workers
in  Flash).

* Make an UIComponent decoupled implementation based on composition over inheritance (here
experience of Alex Harui and other will be very wellcome to start with a good foundation).


* Take into account the SWF and HTML5 outputs in the first drafts.

This would start as an experiment based on fun of coding, and we could see where it goes over
time. If it gets momentum, people join the cause, and so on...

Carlos Rovira
Director de TecnologĂ­a
M: +34 607 22 60 05
F:  +34 912 35 57 77

View raw message