Quoting Erik de Bruin <e...@ixsoftware.nl>:
Mike,
I try to talk about FalconJx, as I believe you call your "version" of
the AS -> JS compiler when I talk about your work.
Ok, I understand. I'm trying to stay away from the phrase of "my way".
I just see it as an alternative that I understand. Which means I can
actually develop it.
I refer to FlexJS
when I talk about the project to allow the user to write in an AS
based framework (either based on Alex's approach or a new one, based
on my experiences setting this up).
Am I correct in assuming that the FalconJx compiler will take any AS
files and translate it to JS using the rules we're working on in the
Wiki?
Word, that is exactly what it does. Not to mention it will also
generate valid AS3 code from the AST as well. I'm actually writing it
against AS3 code generation first because it is the way I see getting
to most precise and comprehensive tests. Once it generates AS3 and is
tested, we override what we need for JS emitters and test that
separately.
I mean, I can build a framework in AS (like Alex's, or maybe
even use Flex itself) and build a complimentary framework in JS (like
Alex did) and as long as the functionalities of these frameworks
match, the user code can be put through FalconJx and the output should
run, correct? How about MXML?
Yes, that is correct. I am doing no black majic here. We are just
setting the foundation and rules for the language translations. Like I
said, having the compiler be able to spit out compilable AS3 code
first means the JS translation and how solid the compiler will be
increases a bunch in my opinion.
I know, maybe others would just jump straight to JS output but, not
me. I have to much experience to know that a thorough foundation will
allow us to hit lightspeed down the road. Besides, I have the time
right now to do this right IMHO.
Mike
EdB
On Mon, Dec 17, 2012 at 7:21 PM, Michael Schmalle
<apa...@teotigraphix.com> wrote:
That's a bit up in the air right now I think.
The way I'm writing tests is AS -> JS google closure. But I'm right in the
middle of refactoring the emitter classes.
The tests I'm writing can be run from ant or whatever, just like the Falcon
tests. I just haven't made a build file yet.
Have you looked at the source code for FalconJx and it's tests?
We don't want to duplicate work but I need to write these tests first. All I
know is, once I get these emitters refactored, the JS output is going to be
what the closure compiler wants. So we could just write an extension in the
assertOut() method that maybe could tests with Jasmine as well? I don't know
I have never used Jasmine testing.
PS, my days with FalconJS are numbered, I've decided it's to complicated for
what I see a relatively simple solution for just getting plain JS, so I
probably won't be working on that code base at all, I think Alex is doing
something with the code.
Mike
Quoting Erik de Bruin <e...@ixsoftware.nl>:
Hi,
I'm trying to figure out how to build tests for the AS to JS cross
compilation.
Should we write AS "strings" and have a script put those through the
FalconJS/FalconJx compiler and compare the result with a pre-defined
"intermediate JS" result? If so, what would be the best tool to write
those tests in? Ant? Shell script? Something else?
I'll be writing the JS tests in Jasmine (mainly because that's what I
know), to test the result from putting the FalconJx output through the
Google Closure Compiler and later on the Closure Builder.
Preferably we would be able to integrate both test suites, so we can
test AS -> "final JS" in one go.
Thoughts?
EdB
--
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
--
Michael Schmalle - Teoti Graphix, LLC
http://www.teotigraphix.com
http://blog.teotigraphix.com
