[antlr-interest] Yet another idea for a completlygenericTreeParser

Loring Craymer Loring.G.Craymer at jpl.nasa.gov
Mon May 16 23:36:41 PDT 2005


At 10:14 PM 5/16/2005, Scott Stanchfield wrote:
>Loring, I'm not getting into this argument with you again. Like I said, we
>obviously don't agree and it's a waste of everyone's time.
>
>People have heard both of our arguments (and know I'm right ;) ).
>
>If this doesn't get into ANTLR I'll have to do it myself, because there's no
>way I'm going through the overhead of tons of extra nodes or object pooling
>to avoid the extra nodes if I can just have a single model object.

Scott--

As I have pointed out several times, "tons of extra nodes" is not what the 
analysis shows.  If you do not have sufficient need for processing 
non-ANTLR trees to experiment with a mechanism which can be effectively 
used now, why do you expect to have sufficient need later to justify the 
effort of maintaining a modified version of ANTLR 3?

Basically, you have gotten into the "designer fools gold" trap.  (I've been 
there.)  Solving the core problem--using ANTLR tree grammars on non-ANTLR 
trees--has become secondary to the "one true solution".  Frankly, I would 
be quite happy to see an example of an ANTLR tree grammar being used to 
process non-ANTLR trees productively and I suspect that others would, 
too.  If you have a good reason for doing this, why not test to see if the 
idea really makes sense?

--Loring


>End of line.
>
>-- Scott




More information about the antlr-interest mailing list