Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!usc!sdd.hp.com!ucsd!ucbvax!agate!bionet!arisia!roo!xerox!barbarin
From: barbarin@Xerox.com (Mike Barbarino)
Newsgroups: comp.lang.c++
Subject: TC++ problem? cont.
Message-ID: <680@roo.UUCP>
Date: 18 Dec 90 21:55:45 GMT
Sender: news@parc.xerox.com
Reply-To: barbarin@Xerox.com (Mike Barbarino)
Organization: Xerox PARC
Lines: 36

Sorry I messed up my previous posting.  I should have posted the actual source rather
than hand-copy from the PC display.  I apologize for wasting the time of all who 
replied...

Here is the test program.  It was compiled under huge and large memory models.  I
initially had DBSIZE set to 30000 (just to see if I could access past 64K).  Since the
problem occurs even when DBSIZE is small, I reset DBSIZE to 30 so that I could do
my debugging in the IDE.

The problem is this:  "weight" doesn't seem to get set correctly, the "for" loop bails
out immediately (more or less).  If I swap "hits" and "weight" in the struct, the
loop finishes correctly.  What's the reason for the inconsistency?

#include <alloc.h>
#include <iostream.h>

main(){
   struct weights {
      long  hits;
      long  weight;
      };

   weights huge *accumulator;

   const long DBSIZE  = 30L;
   const long MEMSIZE = DBSIZE * sizeof(weights);

   accumulator = (weights huge*)farmalloc(MEMSIZE);
   for (int j = 0; j < DBSIZE; j++, accumulator++) {
      accumulator->weight = 255;
      if ((accumulator->weight) != 255) {
         cout << "WHAT HAPPENED AT " << j << endl;
         break;
      }
   }
}