Ok, this is not a parser problem but I didn't see a more suitable component in bugzilla. Anyhow... nsString("foobar").Compare("foo"); returns 0, which is wrong, it doesn't check past the end of the shorter string so that's why it doesn't work. Here's a patch that fixes this problem... diff -c nglayout-latest/mozilla/base/src/nsString.cpp vmproj/mozilla/base/src/nsString.cpp *** nglayout-latest/mozilla/base/src/nsString.cpp Wed Nov 25 13:11:56 1998 --- vmproj/mozilla/base/src/nsString.cpp Fri Nov 27 18:49:05 1998 *************** *** 1534,1542 **** return 1; } if (aIgnoreCase) { ! return nsCRT::strncasecmp(mStr,aCString,maxlen); } ! return nsCRT::strncmp(mStr,aCString,maxlen); } if (aIgnoreCase) { --- 1534,1542 ---- return 1; } if (aIgnoreCase) { ! return nsCRT::strncasecmp(mStr,aCString,maxlen + 1); } ! return nsCRT::strncmp(mStr,aCString,maxlen + 1); } if (aIgnoreCase) { *************** *** 1561,1569 **** return 1; } if (aIgnoreCase) { ! return nsCRT::strncasecmp(mStr,S.mStr,maxlen); } ! return nsCRT::strncmp(mStr,S.mStr,maxlen); } /** --- 1561,1569 ---- return 1; } if (aIgnoreCase) { ! return nsCRT::strncasecmp(mStr,S.mStr,maxlen + 1); } ! return nsCRT::strncmp(mStr,S.mStr,maxlen + 1); } /** *************** *** 1588,1596 **** return 1; } if (aIgnoreCase) { ! return nsCRT::strncasecmp(mStr,aString,maxlen); } ! return nsCRT::strncmp(mStr,aString,maxlen); } if (aIgnoreCase) { --- 1588,1596 ---- return 1; } if (aIgnoreCase) { ! return nsCRT::strncasecmp(mStr,aString,maxlen + 1); } ! return nsCRT::strncmp(mStr,aString,maxlen + 1); } if (aIgnoreCase) {

The problem was that we were calling strncmp rather than strcmp, with a length specified as the longest string in comparision. This caused a match since we only compared exactly n chars.

qa contact set to reporter for verify

http://lxr.mozilla.org/seamonkey/source/xpcom/ds/nsStr.cpp#599 PRInt32 nsStr::StrCompare(const nsStr& aDest,const nsStr& aSource,PRInt32 aCount,PRBool aIgnoreCase) { //Since the caller didn't give us a length to test, and minlen characters matched, //we have to assume that the longer string is greater. is the closest thing i can find to this stuff, it's totally different, but it is designed to consider this issue.