incubator-flex-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jarosław Szczepankiewicz <jszczepankiew...@gmail.com>
Subject Re: Time for an Apache Flex compiler.
Date Thu, 16 Feb 2012 12:49:59 GMT
this is my personal opinion, but what's the purpose of another flex
compiler from open source, especially written in as3?

- will it be faster than mxmlc / fcsh / falcon?
- will it better support multicore architectures / really big projects?
- will it produce better / optimised optocode than mxmlc / falcon ?
- will it better integrate with i.e. syntax higlighing / intellisense
in flash builder?

My opinion is that now we have not ideal but acceptable compiler
(mxmlc). I know that it has some memory leaks, produces not hiper
optimised code, but it works. In one year timeframe we will have far
better support from falcon. In my personal opinion we should wait for
falcon donation and then improve it. There are more imporant long term
targets, especially writing next generation flex that allows
generating html5 output to target ipad / iphone and have 90 percent
coverage in modern browsers without bad marketing imact from flash
player. The FalconJS team published current limitations of flex 4 ->
html5 output generation. This gives us knowledge on now to design
future flex internals to allow hiperoptimized html5 output. Of course
this is apache flex project and and everybody can do anything usable
to apache flex.

2012/2/16 Martin Heidegger <mh@leichtgewicht.at>:
> On 16/02/2012 21:25, Rui Silva wrote:
>>
>> Though I'm generally available for any work that this group decides to
>> undertake, I really do not see this as a viable solution. Do you truly
>> believe that we (considering most of us are Flex developers with little or
>> no expertise in compiler development) could come up with a better solution
>> in less time than a team of professional compiler developers who have been
>> working on it for some time now (even considering that this team has been
>> drastically reduced)? If so, I'm obviously available to help out in any way
>> I can. Best, Rui
>
>
> Writing "a" compiler is easy, writing the flex compiler with all features is
> fairly difficult. However: There is one thing that the new Falcon compiler
> will not solve. It will not be written in AS3 or bytecode and therefore not
> be available directly in as3. I find that sub-optimal which is why I think
> its would be awesome to have one available written in AS3. If you reduce the
> amount of features that this compiler can deal with (for example:
> Compilation of Fonts, jps etc. ) then I see it possible to have one earlier
> than falcon. If its written in a way that it could leverage threads in a
> future version of flashplayer (or workers in JavaScript) then I see no
> reason why this should be significantly slower (at least for most
> operations). If we happen do some things more efficient by luck ... well
> then we are a lucky bunch I guess.
>
> yours
> Martin.

Mime
View raw message