[antlr-interest] Re: Nondeterminism problem

[sarah2geller]

--- In antlr-interest at yahoogroups.com, "Oliver Zeigermann" 
<oliver at z...> wrote:
> --- In antlr-interest at yahoogroups.com, "Weida Ma" <weida_ma at y...>
> wrote:
> > I am getting a nondeterminism warning with a parser:
> > 
> > definition
> > 	:	class_decl_or_def | interface_decl_or_def
> > 	;
> > 
> > class_id
> > 	:	"class" IDENTIFIER
> > 	;
> > 
> > class_decl_or_def
> > 	:	(local_qualifier)? class_id
> > 	;
> > 
> > interface_id
> > 	:	"interface" IDENTIFIER
> > 	;
> > 
> > interface_decl_or_def
> > 	:	(local_qualifier)? interface_id
> > 	;
> > 
> > local_qualifier
> > 	:	"local"
> > 	;
> > 
> > 
> > SliceB.g:38: warning:nondeterminism between alts 1 and 2 of block
> upon
> > SliceB.g:38:     k==1:"local"
> > SliceB.g:38:     k==2:IDENTIFIER
> > 
> > I think "class" and "interface" should be able to distinguish 
> > class_decl_or_def and interface_decl_or_def when k == 2. But it
> seems 
> > I am wrong? I'd appreciate any help.
> 
> This is the prediction code for rule definition:
> 
> if ((LA(1)==LITERAL_class||LA(1)==LITERAL_local) &&
> (LA(2)==LITERAL_class||LA(2)==IDENTIFIER)) {
>   class_decl_or_def();
> } else if ((LA(1)==LITERAL_interface||LA(1)==LITERAL_local) &&
> (LA(2)==IDENTIFIER||LA(2)==LITERAL_interface)) {
>   interface_decl_or_def();
> }
> 
> This actually corresponds to the above warning. It seems to me you 
are a victim of approximated LL lookahead :( 
> 
> Or am I wrong? Experts?
> 
> Oliver

Correct. The grammar is strong LL(2), confirmed by the SLK parser-
generator (http://parsers.org).




 

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