[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Gnash-dev] address@hidden: [osflash] Lightspark preliminary support

From: Alessandro Pignotti
Subject: Re: [Gnash-dev] address@hidden: [osflash] Lightspark preliminary support for Flex and the Windows platform]
Date: Wed, 27 Jan 2010 18:16:24 +0100
User-agent: KMail/1.12.4 (Linux/2.6.30-2-amd64; KDE/4.3.4; x86_64; ; )

Hi strk and everyone in the list,

thanks for CCing me. I've been following gnash-dev mailing list for a long 
time, but I've never contributed to the discussion. Yes, I remember that you 
proposed me to join the gnash project, and I actually spent some time looking 
at the gnash's AVM2 implementation, before resuming the work on my own 

My point is that is very difficult to include an optimizing JIT compiler inside 
an existing architecture, the main advantage of working on Lightspark was to 
have a clean design, build from the ground up to support JIT compilation and 
accelerated rendering. My JIT, in the current status, exhibit performances 
that are just a bit behind tamarin on benchmarks included in the tamarin 
source, but there is a lot of room for improvements. Moreover, as I'm using 
LLVM as the backend, Lightspark will automatically take advantage of the work 
on such framework.

I have the maximum respect for the result of the gnash project, as you started 
working when no specs were available, and that's definitely an impressive work. 
Moreover, gnash supports to a large extend AVM1 for Actionscript 1.0 and 2.0, 
and I have no plan to support them in Lightspark, as I expect people to 
migrate to AS3 over time, especially for computationally intensive tasks. I 
think that to some extend Lightspark and Gnash are going to share part of the 

I warmly invite people in the gnash community that may be interested in 
joining Lightspark to take a look at the code and to send me any question 
about architectural issues or anything else. I'm not a professional software 
designer and, although I'm finding the current structure pretty scalable, I'm 
sure that there is a lot of room for advices from people more expert than me.

Thanks a lot,

On Wednesday 27 January 2010 12:52:22 strk wrote:
> FYI.
> I can't remember if we talked with Alessandro already in the past about
> joining our efforts.  Alessandro, do you remember about it ?
> ----- Forwarded message from Alessandro Pignotti <address@hidden>
>  -----
> Date: Tue, 26 Jan 2010 05:01:54 +0100
> From: Alessandro Pignotti <address@hidden>
> Subject: [osflash] Lightspark preliminary support for Flex and the Windows
>       platform
> Reply-To: Open Source Flash Mailing List <address@hidden>
> To: address@hidden
> X-BeenThere: address@hidden
> Hi everyone,
> maybe someone still remember Lightspark, my personal struggle to create an
> open source flash player. Well, although I've not been actively publishing
>  news about it for several months now, work has been going on full speed
>  ahead. As part of my bachelor thesis I've greatly enhanced and tested my
>  Actionscript 3.0 JIT engine based on LLVM. For the last week or so I've
>  been working to make the Virtual Machine stable enough to run stuff based
>  on the Flex framework. The test application I've been working on is mostly
>  empty, just a bare mx.core.Application with a VBox children, generated
>  using the following mxml source.
> <?xml version="1.0" encoding="utf-8"?>
> <mx:Application  xmlns:mx="";
>     horizontalAlign="center" verticalAlign="middle">
>       <mx:VBox x="0" y="0" width="201" height="200" backgroundColor="0x0080C0"
> alpha="0.8"/>
> </mx:Application>
> This may appear trivial, but I can assure you that a huge deal of work is
>  done by the flex framework to run this.
> In some days I'll release a new technical demo but, as always, code is
>  already available using git from the repository
> git://
> I think that having the Flex framework running is a great sign that the
> Virtual Machine is stable enough to move the project from a pre-alpha to an
> alpha status. Moreover, now I think the design itself is good enough for
>  other people to join the effort, as I'll never be able to finish the work
>  all by myself. I'm looking for people with good C++/OpenGL programming
>  experience. With this release preliminary support for Windows has been
>  added, so people from both the Windows and Linux world may join.
> Right now there are some major stuff to work on:
> -) Write an Actionscript interpreter: since the beginning lightspark was
>  based on a JIT engine to run Actionscript, as that was also the main focus
>  of my thesis. But actually is very sensible to set up an interpreter to
>  run code that is not used often, such as initialization code. Right now
>  the whole system is pretty slow, but more than 90% of the time is spent
>  inside LLVM Just in time compiler. Moreover, most of the code is compiled
>  and then executed only once, which results in a waste of time and memory.
> -) Optimize, and maybe redesign, the graphics rendering system. I'm not an
> OpenGL guru (although definitely not a newbie), so advices from people more
> expert than me would be pretty useful.
> -) Make the Actionscript engine pass all the conformance tests included
>  with the tamarin source code
> -) Add support for Actionscript exceptions. That's the only major feature
>  of the language still totally unsupported.
> Cheers,
> Alessandro Pignotti
> _______________________________________________
> osflash mailing list
> address@hidden
> ----- End forwarded message -----

Attachment: signature.asc
Description: This is a digitally signed message part.

reply via email to

[Prev in Thread] Current Thread [Next in Thread]