[antlr-interest] Re: literals.

lgcraymer lgc at mail1.jpl.nasa.gov
Wed Dec 10 13:32:18 PST 2003


Check the rules which call relational_op as well:  it's more likely that th=
e source of the (apparent) nondeterminism is there, but that 
the FOLLOW set is being evaluated within relational_op.  Sometimes the nond=
etermism warnings are due to the approximate LLk, but 
other times they indicate a problem and the warnings may tell you where the=
 nondeterminism was detected but not where it is 


--- In antlr-interest at yahoogroups.com, cesar octavio lopez nataren <cesar at c=
...> wrote:
> On Wed, 2003-12-10 at 04:41, Cristian Amitroaie wrote:
> > Why not simply adding to the lexer:
> > 
> > tokens {
> >    IN="in";
> >    INSTANCE_OF="instanceof";
> > };
> Hello Cristian!
> dude! this solution rocked! thanks a lot. I get the nondeterminism
> warning with this, but it's picking the correct rule at each case. So
> I'll ignore that warning ;-)
> By the way, I tried Oliver suggestion about IN taking precedence over
> IDENTIFIER, it did work, but after that I could not recognize
> identifiers that were having in as prefix: inverse;
> 			                   ^^
> Thanks a lot for the suggestions guys, you rock!
> César


Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/ 

More information about the antlr-interest mailing list