From: Mark Phillips <umphill5 AT cs DOT umanitoba DOT ca>
Newsgroups: comp.os.msdos.djgpp
Subject: destructor problem
Date: Fri, 28 May 1999 13:39:17 -0500
Organization: The University of Manitoba
Lines: 42
Message-ID: <Pine.GSO.3.96.990528132952.18715A-100000@gold.cs.umanitoba.ca>
NNTP-Posting-Host: gold.cs.umanitoba.ca
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
X-Trace: canopus.cc.umanitoba.ca 927916766 6300 130.179.24.1 (28 May 1999 18:39:26 GMT)
X-Complaints-To: Postmaster AT cc DOT umanitoba DOT ca
NNTP-Posting-Date: 28 May 1999 18:39:26 GMT
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Reply-To: djgpp AT delorie DOT com


hi.  i have a destructor that just calls a close member function:

foo::~foo()
{
   close();
}

close() causes a gpf when the destructor calls it.  on the other hand if i
comment out the call to close, and explicitly call it when an instance
goes out of scope like so:

int main()
{
   foo instance;
   ...
   instance.close();
   return 0;
}

it executes normally.  that seems pretty weird to me.  anyone have any
ideas?  i don't see why close() would cause a gpf in the first place
though, it seems pretty safely coded:

void foo::close()
{
   if(data!=NULL)
   {
      unlock();
      delete [] data;
   }
   data=NULL;
   curr=NULL;
   size=0;
}

by the way, i traced the gpf back to the 'delete [] data' line.  even when
i explicitly call the close() function multiple times in a row right
before going out of scope it works fine.  TIA

mark