[antlr-interest] problem with nondeterminism

[Michael Wirth]

Thanks, now I can see the problem too.

Am Samstag, 10. Mai 2003 00:10 schrieben Sie:
> When antlr analyzes (unit)*, it sees that PROGRAM javaconstruct could be
> parsed two ways.  It's a lot like the if/else ambiguity.  You can turn off
> the warning with the warnWhenFollowAmbig=false option.  See the docs about
> it.
>
> Monty
>
> -----Original Message-----
> From: Wirth. Michael
> To: antlr-interest at yahoogroups.com
> Sent: 5/8/2003 3:21 AM
> Subject: [antlr-interest] problem with nondeterminism
>
> Hello,
>
> here is the code.
>
> -----------------------------------------
> start: (unit)* ;
>
> unit : program
>
>      | procedure
>      | globaljavastart
>
>      ;
>
> program : PROGRAM (statement)* ;
>
> statement :
>           javaconstruct
> //        | "if" ...
> //        | "while" ...
> //        ...
>           ;
>
> javaconstruct :
>         javastart
>
>       | globaljavastart ;
>
> globaljavastart :
>         GLOBALJAVASTART GLOBALJAVAEND ;
>
> javastart :
>         JAVASTART JAVAEND ;
>
> procedure :
>         PROCEDURE (statement)* ;
> -----------------------------------------
>
> antlr generates the folloing warnings
> (for the rule "program" and "procedure")
>  warning:nondeterminism upon
>      k==1:GLOBALJAVASTART
>      between alt 1 and exit branch of block
>  warning:nondeterminism upon
>      k==1:GLOBALJAVASTART
>      between alt 1 and exit branch of block
>
> I think the problem is that the rule "globaljavastart"
> can be used in a statement and in a unit.
> Can I define the grammar in an other way?
> Or can I ignore these warnings?
>
> I hope someone can help me, because I need this
> in my diploma thesis.
>
> Greetings,
> 	Michael
>
>
>
> Your use of Yahoo! Groups is subject to
> http://docs.yahoo.com/info/terms/


 

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