[antlr-interest] Access Violation in InitLiterals()

[Kevin J. Cummings]

Hi Folks,
	I writing my 3rd of 4 grammars using ANTLR 2.7.7.  The first 2 are
done, and because they don't involve any IDENTs, all Keywords were
defined in the Lexer with TOKEN names.  In my 3rd and 5th grammars, the
Keywords must co-exist with the IDENTs, so I'm using literals in my
grammar to avoid the non-determinism between my IDENTs and the keywords.

	I tested this grammar on my Linux system without many actions in either
the Parser or the TreeWalker.  Runs just fine.

	BUT, I am developing this in VC++ on a Windows XP system.  When I
attempted to run it for the first time, it dies, in InitLiterals() on
the following statement (the first assignment in this routine):

> 	literals["ends_with"] = 61;

By following the backtrace, I find myself in the following code in
CharScanner.h:

> inline bool CharScannerLiteralsLess::operator() (const ANTLR_USE_NAMESPACE(std)string& x,const ANTLR_USE_NAMESPACE(std)string& y) const
> {
> 	if (scanner->getCaseSensitiveLiterals())
> 		return ANTLR_USE_NAMESPACE(std)less<ANTLR_USE_NAMESPACE(std)string>()(x,y);
> 	else
> 	{
> #ifdef NO_STRCASECMP
> 		return (stricmp(x.c_str(),y.c_str())<0);
> #else
> 		return (strcasecmp(x.c_str(),y.c_str())<0);
> #endif
> 	}
> }

On the call to stricmp(), the y.c_str() argument appears to be
un-initialized (the value of the pointer is 0xcdcdcdcd).

What really bugs me is why does this work on Linux and not on windows.
My first 2 grammars don't use literals, so this routine is empty in those.

Does anyone have any advice on what I can do to make this work?

Thanks!

-- 
Kevin J. Cummings
kjchome at rcn.com
cummings at kjchome.homeip.net
cummings at kjc386.framingham.ma.us
Registered Linux User #1232 (http://counter.li.org)