From: Al Christians <achrist AT easystreet DOT com>
Newsgroups: comp.os.msdos.djgpp
Subject: Template/STL questions
Date: Wed, 12 Aug 1998 18:54:42 -0700
Organization: Trillium Resources Corporation
Lines: 45
Message-ID: <35D24762.7E27@easystreet.com>
Reply-To: achrist AT easystreet DOT com
NNTP-Posting-Host: dial-35-088.easystreet.com
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
NNTP-Posting-Date: 13 Aug 1998 01:47:00 GMT
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Precedence: bulk

I'm having trouble with a big program failing with little dignity and
a segment fault, and am fishing for any ideas about what might be
breaking it.  TIA for any suspicions or likely suspects.  The program
makes heavy use of the <string>, <map>, <vector>, and a few other
related headers. 

One symptom that looks real baad is that when I do 

string a;

I don't get an empty string, but

string a("");

in the same place does work ok.  So it seems that something is perhaps
clobbering indiscriminately.  This leads to suspicions about all the
templates getting used.  Another thing we find is that

	vector<string>  b(100);
	b.clear()
	b[0] = "0";

will work ok, but

	vector<string>    b;
	b.push_back(string("0"));  

will fail, i.e. any attempt to access b after the push_back gives the
segment error.  But doing it the other way

	b[0] = string("0");

will also cause the failure right there. 	

There are plenty of STL collections allocated on the stack,  and I
fear that the stack might be getting corrupted as these things are
supposed to be growing and shrinking automatically.  Specifically, 
we have some  map<string, vector<string> > objects on the stack where
the vectors within the map are supposed to be growing and shrinking.   

Changing the stack size between 500k and 4M doesn't seem to help at
all.  

 
Al