[antlr-interest] Epsilon productions and end of file

[Arnar Birgisson]

Hello Marcin,

Could you post a portion of your ANTLR grammar?

Btw: Are you eliminating left-recursion by adding F'? If so it's
convenient to use ANTLR-specific syntax instead (the ? or + operators).

Arnar

> -----Original Message-----
> From: Marcin Rzeznicki [mailto:marcin-rzeznicki at wp.pl] 
> Sent: 17. desember 2003 21:44
> To: ANTLR mailing list
> Subject: Re: [antlr-interest] Epsilon productions and end of file
> 
> 
> ----- Wiadomosc oryginalna ----- 
> Od: "Arnar Birgisson" <arnarb at oddi.is>
> Do: <antlr-interest at yahoogroups.com>
> Wyslano: 17 grudnia 2003 12:13
> Temat: RE: [antlr-interest] Epsilon productions and end of file
> 
> > > > This means that you must specifically tell ANTLR where to expect
> > > > end-of-file, and for that you use the special token EOF.
> > >
> > > Yes, but when I use explicit EOF, it does not calculate
> > > follow sets as I
> > > would expect. For example
> > >
> > > S -> AB , means taht follow set of S goes to follow set 
> of B (with $)
> > >
> > > S -> AB EOF ends with EOF in follow set of S, nothing to do with B
> >
> > I'm sorry, I don't quite follow. S -> AB EOF should not 
> result in EOF
> > being in FOLLOW(S), but in FOLLOW(B).
> 
> Of course, I have misplaced symbols. It should be: EOF goes 
> to FOLLOW(B),
> but FOLLOW(S) is not in FOLLOW(B) anymore. I am sorry. What I 
> mean is that
> by writing S->AB I expect FOLLOW(S) to be in FOLLOW(B), as FOLLOW(S)
> contains EOF, so does FOLLOW(B). By writing S->AB EOF it ends 
> up in EOF
> being in FOLLOW(B), but without everything else from FOLLOW(S). In my
> example the grammar is like:
> F->DF'
> D->...
> P->(F) | ...
> Constructing LL parser with pencil I have assumed that ')' is 
> in FOLLOW(F),
> therefore in FOLLOW(F'), and every word of generated lanuage may be
> terminated by ')' or EOF. But parser did not want to choose epsilon
> productions that should occur when EOF was read, but tried to 
> look ahead for
> another token, which resulted in throwing an exception. So it 
> only accepted
> words terminating with ')'
> 
> When I transformed grammar to:
> F->DF' EOF
> ....
> P->(F)|...
> FOLLOW (F) contains ')' as well as EOF, but then FOLLOW(F') 
> does not contain
> ')', so parser does not accept words that have ')' somewhere within.
> 
> That's my problem.
> 
> --
> Greetings
> Marcin Rzeznicki
> 
> 
>  
> 
> Yahoo! Groups Links
> 
> To visit your group on the web, go to:
>  http://groups.yahoo.com/group/antlr-interest/
> 
> To unsubscribe from this group, send an email to:
>  antlr-interest-unsubscribe at yahoogroups.com
> 
> Your use of Yahoo! Groups is subject to:
>  http://docs.yahoo.com/info/terms/ 
> 
> 
> 


 

Yahoo! Groups Links

To visit your group on the web, go to:
 http://groups.yahoo.com/group/antlr-interest/

To unsubscribe from this group, send an email to:
 antlr-interest-unsubscribe at yahoogroups.com

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