Up-to-date documentation

[xolotl]

Is there anything as thorough as the thesis for the current version of
SableCC?

Thanks,

-- O.L.


_______________________________________________
SableCC-Discussion mailing list
SableCC-Discussion@...
http://lists.sablecc.org/listinfo/sablecc-discussion

[Etienne M. Gagnon]

Hi Olivier,

SableCC 3.2 is mostly SableCC 2.x (which is documented in my M.Sc. thesis) with 2 additional features:

1. CST to AST grammar-specified transformations.
2. Generation of Java 1.5+ code (using parametric types for lists).
   

Tree transformations are somewhat described in Komivi (Kevin) Agbakpem's M.Sc. thesis, in French:

• http://sablecc.sourceforge.net/downloads/komivi-kevin-agbakpem-msc-2006.pdf
  

 Yet, I would recommend reading the following document, which provides shorter explanations in English:

• http://nat.truemesh.com/archives/000531.html

As for the parametric types, they are not documented, yet, they are an obvious translation of Java 1.4- into 1.5+.

Hoping this helps,

Etienne

Olivier Lefevre a écrit :

Is there anything as thorough as the thesis for the current version of 
SableCC?
  

_______________________________________________
SableCC-Discussion mailing list
SableCC-Discussion@...
http://lists.sablecc.org/listinfo/sablecc-discussion

[xolotl]

>    1. CST to AST grammar-specified transformations.

These are very similar to ANTLR's rewrite rules, correct?

-- O.L.


_______________________________________________
SableCC-Discussion mailing list
SableCC-Discussion@...
http://lists.sablecc.org/listinfo/sablecc-discussion

[Etienne M. Gagnon]

When Kevin did his work, ANTLR did only have a syntax to retain some tokens/rules in a tree, not to transform the tree. It is possible that ANTLR has added a new syntax, later, as Terrence Par came to visit us at UQAM and we shared a few ideas and had a lot of fun. I haven't kept up to date with its latest development, so I can't assert anything.

By transforming the tree, I mean, changing it such as :

CST
 program = declaration* {-> new program([declaration.variable],[declaration.function])};

 declaration {-> variable, function}=
  {function} function {-> Null, function} |
  {variable} variable {-> variable, Null};

AST
  program =
   variable* function*;

See how the CST allows declarations in any order, yet the AST has variables first.
 
Etienne

Olivier Lefevre a écrit :

   1. CST to AST grammar-specified transformations.
    

These are very similar to ANTLR's rewrite rules, correct?
  

_______________________________________________
SableCC-Discussion mailing list
SableCC-Discussion@...
http://lists.sablecc.org/listinfo/sablecc-discussion

[xolotl]

Etienne M. Gagnon wrote:
> When Kevin did his work, ANTLR did only have a syntax to /retain/ some
> tokens/rules in a tree

Yes, that's the old tree operators. They're still here.

> not to /transform/ the tree. It is possible that ANTLR has added a new syntax

It has. I can't compare in detail but the idea is the same: where operators
won't suffice, rewrite rules let you perform more radical transformations.

-- O.L.


_______________________________________________
SableCC-Discussion mailing list
SableCC-Discussion@...
http://lists.sablecc.org/listinfo/sablecc-discussion