Thanks for answering those questions Gordon.

It sounds like this once settled in could be a tool developers dream come true. I'm really excited about this and I'm sure it will be worth the wait.

Having some experience in this area, with just the information you gave(API and such) I can already start to think about some projects I want to do to extend it or develop tools that execute it and use it's DOM.

Thanks again!
Mike

Quoting Gordon Smith <gosm...@adobe.com>:

Does the ANTLR parser use the standard token stream if it's not lexing the source?

For AS, where we use JFlex to do the lexing (for better performance), we implement our own token stream.

For CSS, where ANTLR is doing the lexing, we use its standard token stream.

Would there be a way to create a linked list token stream with what the lexer creates?

We don't buffer visited tokens for AS, but it would be straightforward to implement your own linked list token stream.

Is there any plan on adding FXG grammar and DOM?

I forgot to mention that Falcon does support compiling FXG. But we didn't innovate much in this area... we mainly just brought over the FXG code from the old compiler, which you can take a look at. I believe it uses a SAX parser to parse the XML of the FXG file, and it builds a custom DOM with classes like RectNode and TextNode which it then uses for semantic analysis and code generation. One improvement is that the FXG compilation takes place in the same multithreaded fashion as other Falcon compilation.

- Gordon

-----Original Message-----
From: Michael Schmalle [mailto:m...@teotigraphix.com]
Sent: Thursday, January 19, 2012 4:48 PM
To: flex-dev@incubator.apache.org
Subject: RE: Falcon compiler source code / Falcon architecture

Gordon, when you have time;

Does the ANTLR parser use the standard token stream if it's not lexing the source?

Would there be a way to create a linked list token stream with what the lexer creates?

Is there any plan on adding FXG grammar and DOM (this would be awesome but maybe something the community could implement)?

Thanks,
Mike


Quoting Gordon Smith <gosm...@adobe.com>:

ANTLR is used to parse (but not lex) AS, and to both lex and parse CSS.

- Gordon.

-----Original Message-----
From: Michael Schmalle [mailto:m...@teotigraphix.com]
Sent: Thursday, January 19, 2012 11:16 AM
To: flex-dev@incubator.apache.org
Subject: RE: Falcon compiler source code / Falcon architecture

Gordon this is about the best news we could get.

It's great to see this structure! I can't wait.

What parts are using ANTLR grammar?

Thanks,
Mike

Quoting Gordon Smith <gosm...@adobe.com>:

Until we release specs, Javadoc, or code for Falcon, here is a very
brief overview of Falcon's architecture:

Falcon is designed from the beginning to support compiling multiple
targets in multiple projects in a workspace (as you have in an IDE
like Flash Builder). By contrast, asc was designed to compile a
single AS file, and then mxmlc was built on top of that.

Falcon uses multiple threads to compile multiple files at the same
time. The more cores you have, the faster it goes.

Critical data structures such as the symbol table (which stores
information about which classes are known, what methods they have,
etc.) are shared across the entire workspace, to minimize memory
usage.

Critical data structures are maintained in memory to support both
compilation and IDS code intelligence in an efficient and consistent
way. For example: If you open a file in the IDE, Falcon builds a
syntax tree and symbol table for it to support intelligent editing.
Compiling the file requires just one additional code generation step.
By contrast, in the current Flash Builder which uses the compiler in
the SDK, Flash Builder builds its own parse trees and symbol tables
to support editing, and then when you compile the compiler in the SDK
builds another set of parse trees and symbol tables. This is slow and
a waste of memory.

Falcon understands .as, .mxml, .css, and .properties files.

The parse trees for AS consist of nodes from about 100 classes, such
as LiteralNode, BinaryOperatorNode, FunctionNode, ClassNode. The
parse trees for MXML consist of nodes from about 50 classes, such as
MXMLDocumentNode, MXMLInstanceNode, MXMLScriptNode, etc. The
ActionScript-y parts of MXML are represented by AS nodes inside of
MXML nodes.

The symbol table consists of objects representing the things your
code defines, such as ClassDefiniition, FunctionDefinition, and
VariableDefinition, arranged into a hierarchy of scope objects.

Falcon uses 3rd-party grammars like JFlex, ANTLR, and JBurg to
generate various lexers, parsers, and code generators.

MXML is compiled directly to ABC, not to ActionScript source code or
an ActionScript parse tree.

- Gordon Smith, Adobe

P.S. After working on the Flex framework since its inception for
about
8 years, I joined the Falcon team 18 months ago.



-----Original Message-----
From: Raju Bitter [mailto:rajubit...@googlemail.com]
Sent: Thursday, January 19, 2012 2:10 AM
To: flex-dev@incubator.apache.org
Subject: Re: Falcon compiler source code / Falcon architecture

2012/1/19 Alex Harui <aha...@adobe.com>:
I was in the Falcon code in November.  It was way easier to figure
out how to make changes than with MXMLC.  Even if it is not
documented, I think it will be easier to make progress that with the
Falcon code base.
I agree, that sounds like it would be much better to use the Falcon
code base. There are normally fewer community members with much
experience in language/compiler design, and the cleaner the
architecture is, the better for the community.









Reply via email to