[antlr-interest] CommonTree & Tree grammar versus DIY

Terence Parr parrt at cs.usfca.edu
Sat Aug 16 12:30:47 PDT 2008


On Aug 16, 2008, at 12:56 AM, Gerald Rosenberg wrote:
> Even with so much existing, this is not a trivial extension to  
> Antlr.  Theoretically, the end result of any "doSomething()" AST  
> manipulation could be done with a well-planned set of tree-walkers.   
> It is just that a random access approach, supported by the  
> equivalent of findFirst, findNext, findPrev, findLast operations, is  
> distinctly better than a purely top-down (only getNext allowed)  
> approach for non-trivial bi-directionally context-dependent AST  
> rewrite problems.
>
> Unfortunately, does not sound like AntlrMorph addresses this  
> particular problem.

Not sure I understand the specific problem; can you rephrase?

> Ter, do you have another graduate student looking for a project?   
> Or, if I finish my implementation, can I get Ph.D. credits? ;)

sure ;)
Ter


More information about the antlr-interest mailing list