[antlr-interest] [3.1 C] Mulivalue Attribute is NULL in action after matching rule

[Jim Idle]

No - it is just the lack of output=AST;

Jim

> -----Original Message-----
> From: antlr-interest-bounces at antlr.org [mailto:antlr-interest-
> bounces at antlr.org] On Behalf Of Gavin Lambert
> Sent: Saturday, April 12, 2008 7:19 PM
> To: Christian and Natalie Schladetsch
> Cc: antlr-interest at antlr.org
> Subject: Re: [antlr-interest] [3.1 C] Mulivalue Attribute is NULL in
> action after matching rule
> 
> At 13:48 13/04/2008, Christian and Natalie Schladetsch wrote:
> >grammar Fail;
> >options { language=C; }
> >@preincludes { #include <assert.h> }
> >sequence  : num += (NUMBER+) { assert($num != 0); } ;
> >NUMBER  :  ('0'..'9')+ ;
> 
> That other problem (with parentheses) has crept back in :)
> 
> >error C2039: 'vectors' : is not a member of 'PiParser_Ctx_struct'
> >
> >That is the complete parser. The line in question is
> >'list_num=ctx->vectors->newVector(ctx->vectors);'
> >
> >With regards of mismatched templates to runtime: I fear that the
> >work-around I used to get antlr built (by deleting files) results
> >in this. So my bad. I guess I need to go back to the drawing
> >board and make a new build of antlr 3.1 somehow.
> 
> Are you building your grammar with ANTLRworks or via the command
> line?  ANTLRworks contains embedded copies of all the ANTLR
> classes and templates, which you'll need to replace with the 3.1
> versions if you want to use the 3.1 runtime.
> 
> (I usually don't bother; if I need to use an intermediate build
> for some reason [and I've always used the prebuilt ones] then I'll
> just stop using ANTLRworks until it catches up.  Since I'm not
> targeting Java and it doesn't properly support Unicode input
> anyway, I've only found it of limited usefulness.)
> 
> Another thing to check is your include path; maybe it's finding
> one of the other versions of the C runtime before the one you
> actually want to use.