[antlr-interest] Exceptional question...:)

[Hill, Robert]

That's the ticket Martin, cheers.
I think having the predicate in there narrowed my vision, and I didn't
want to remove it for some unknown reason.!
I've added the test after the match, and it works just fine.

Also about the ##, I didn't think about the cut and pasting of code, I
guess that depends on how many AST type statements you have in the rule
:)

Cheers anyway :)

/2ob
 
-----Original Message-----
From: antlr-interest-bounces at antlr.org
[mailto:antlr-interest-bounces at antlr.org] 
Sent: 20 June 2005 11:23
To: Robert Hill
Cc: antlr-interest at antlr.org
Subject: Re: [antlr-interest] Exceptional question...:)

Hi,

just match the characters and throw the exception afterwards:
> 
> Element
> 
> Options { defaultErrorHandler = false; }
> 
> :
> 
>       ty:type nn:ID (ID! | HYPHEN!) cnt:element_count (INT | HYPHEN) 
> string_size[#ty.getText()]
> 
>             (INT | HYPHEN)  ({bIgnoreExtraneousEOLChars}? (options
> {greedy=false;}:~CR!)* CR! 
            (INT | HYPHEN)  ( (options {greedy=false;}:~CR!
	{ if (!bIngoreExtraneouseEOLChars) throw ...; } )* CR!

>                   ## = #([TOK_COMMENT,t]);
> 
>                   #comment = #([TOK_COMMENT,t]);
> 
>                   #comment = #(#[TOK_COMMENT,t]);

You can also write ## = #(#[TOK_COMMENT],#t); ;-)

I think it's just some liberation from the syntax, the idea is that you
don't have to write all those # chars if you're in an AST construction
statement anyways. About the ##/#comment: most people seem to prefer ##
as it's shorter and also doesn't break if you move the code or rename
the parser rule.

Regards,
Martin