Quoting Alex Harui <[email protected]>:
Last I looked, you were still using the threads to get the bytes from each
compilation unit, which is good. Yesterday, I started in on CSS support in
FlexJS and realized that the CSS support in Falcon also uses the BURM. Are
you going to re-write that and any other sub-compilers as well (resources,
fxg)?
Threads, yes I fully understand the framework now and its Futures
implementation (request tasks). I also know that I am using
multi-threading in full parse but not in source code emitting which I
am not concerned about since its pretty fast just producing
strings(verses SWF bytecode reduction).
That said, in the future I know how to the get source code production
multi-threaded as well, that is just an implementation step.
As far as CSS, how do you handle that as far as a production?
FXG would be just another traverse on the DOM, but what are you
talking about with properties parsing? Why would we have to rewrite
the properties parser?
Mike
-Alex
On 3/6/13 12:18 PM, "Michael Schmalle" <[email protected]> wrote:
Alex,
The business logic side of things I am sure is pretty close to what
you have in FalconJS.
When Erik finishes the MXML, there is no reason to even think about
FalconJS anymore. The jx framework is completly capable of creating
any type of output we want and also extends itself to others in the
community openly creating NEW output formats to suit there own needs.
This framework is truly in the spirit of Apache and will be supported
for some time to come.
I am looking forward to Erik getting the MXML working so we can test
more real world AS against the emitters and find any weird bugs.
Mike
Quoting Alex Harui <[email protected]>:
OK. I'm still hoping to be able to take a closer look at what you guys are
doing in the next couple of weeks.
On 3/6/13 11:57 AM, "Michael Schmalle" <[email protected]> wrote:
I'm just replying for the sake of completeness.
All I said today was I found an original impl basically reusing the
logic you had in FalconJS.
Erik is working on the MXML, so I am not even getting close to anybody
thinking I am trying to step on feet, I should have just said nothing
and deleted the file. :)
I'm working on other todos that are getting the base a lot more stable
so I can leave it for a time. The core visitor framework is stable in
my opinion anyway.
Mike
Quoting Alex Harui <[email protected]>:
Trying to catch up on this thread. Instead of replying to individual
messages let me see if I can summarize my input here:
The MXML code in Falcon is pretty much doing a tree walk, so I
don't quite
get why it can't be mostly re-used.
The mainstream Flex framework has code that understands the data
structures
so in theory you don't need to do other output for Vanilla
We can adjust the data structures if Gordon's concerns are
important, but I
don't think they are critical at this time.
On 3/6/13 11:12 AM, "Michael Schmalle" <[email protected]> wrote:
Well that's positive news.
My experience came from trying to understand what Alex was doing in
the rats nest code of FalconJS.
So if its a lot easier than I fist imagined that is great then.
Mike
Quoting Erik de Bruin <[email protected]>:
Stepping through the current FlexJS MXML with "my" parser, I am
confidant that I can create the needed data structures using the setup
that is currently in the code. If I'm somehow mistaken, we'll know
soon enough.
EdB
On Wed, Mar 6, 2013 at 8:01 PM, Michael Schmalle
<[email protected]> wrote:
No,
I said there is a huge curve to what you are about to try and
produce with
data structures. I'm asking have you thought about how your
are going to
create them... ?
His data structures are pretty freaking abstract and you
can't just loop
through the DOM and create them.
Mike
Quoting Erik de Bruin <[email protected]>:
Mike,
Aren't we crossing streams now? My efforts are aimed at creating
FlexJS output and my code is close to making that happen. If I
understand your last correctly, you are suggesting a different
approach?
EdB
On Wed, Mar 6, 2013 at 7:54 PM, Michael Schmalle
<[email protected]> wrote:
Alex,
Did you happen to read Gordon's comments on
https://cwiki.apache.org/confluence/display/FLEX/MXML+Data+Spec
At the end of Jan?
I'm seriously thinking about giving an initial impl for
FalconJx right
now
and your data structures.
I just found the 1400 line prototype emitter/walker I made
a month ago
dealing with this and your FragmentList.
Since I'm all about getting this boat shoved off, I can
afford a couple
days
trying to get what I started working abit.
Mike
Quoting Michael Schmalle <[email protected]>:
Hmm.
I understand what you are saying but that is what the walker is for
because most output is never going to translate to the existing
structure of
what will be output as the case with FlexJS and it's data
structures. In
AS
we have a 1 to 1 relationship with the AST to source code.
I see this as duplicate work considering in MXML you are basically
cherry
picking what you need and creating an internal model based
on the data
parsed which then you will loop through and create the js class.
But, that is just the way I see it. :)
Mike
Quoting Erik de Bruin <[email protected]>:
Mike,
The idea is to create the same "input == output" tests for
MXML as you
did for AS, to check if the emitters are handling all input
correctly.
Inheriting from those verifiably 'correct' MXML emitters I
will create
JS output, both for FlexJS and VanillaSDK (initially, but
others might
be added later), similar to how we do both AMD and 'goog'
JS output
from extending the AS emitter class. MXML in my view is a
second input
type, next to AS.
EdB
On Wed, Mar 6, 2013 at 7:30 PM, Michael Schmalle
<[email protected]> wrote:
Erik,
I'm just looking at some of the things you have been working on.
I'm a bit confused... My understanding was the MXML emitters
needed to
produce relevant AS code of FlexJS IE Alex's data structures.
Am I missing something obvious here that was discussed
between you and
Alex?
When I created the base walker and emitter I never had
intentions of
producing MXML source code from the MXML AST DOM.
Mike
--
Michael Schmalle - Teoti Graphix, LLC
http://www.teotigraphix.com
http://blog.teotigraphix.com
--
Ix Multimedia Software
Jan Luykenstraat 27
3521 VB Utrecht
T. 06-51952295
I. www.ixsoftware.nl
--
Michael Schmalle - Teoti Graphix, LLC
http://www.teotigraphix.com
http://blog.teotigraphix.com
--
Michael Schmalle - Teoti Graphix, LLC
http://www.teotigraphix.com
http://blog.teotigraphix.com
--
Ix Multimedia Software
Jan Luykenstraat 27
3521 VB Utrecht
T. 06-51952295
I. www.ixsoftware.nl
--
Michael Schmalle - Teoti Graphix, LLC
http://www.teotigraphix.com
http://blog.teotigraphix.com
--
Ix Multimedia Software
Jan Luykenstraat 27
3521 VB Utrecht
T. 06-51952295
I. www.ixsoftware.nl
--
Alex Harui
Flex SDK Team
Adobe Systems, Inc.
http://blogs.adobe.com/aharui
--
Alex Harui
Flex SDK Team
Adobe Systems, Inc.
http://blogs.adobe.com/aharui
--
Alex Harui
Flex SDK Team
Adobe Systems, Inc.
http://blogs.adobe.com/aharui
--
Michael Schmalle - Teoti Graphix, LLC
http://www.teotigraphix.com
http://blog.teotigraphix.com
