Xref: utzoo gnu.gcc:1242 gnu.gcc.bug:1829
Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!cs.utexas.edu!samsung!think!snorkelwacker!bloom-beacon!shelby!mcnc!thorin!unc!alexande
From: alexande@unc.cs.unc.edu (Geoffrey D. Alexander)
Newsgroups: gnu.gcc,gnu.gcc.bug
Subject: Possible gcc bug in sscanf
Keywords: GNU C, sscanf, bug
Message-ID: <11393@thorin.cs.unc.edu>
Date: 10 Jan 90 17:46:48 GMT
Sender: news@thorin.cs.unc.edu
Lines: 39

I have encountered what I believe to be a GNU C bug in sscanf.  The problem is
illustrated by the following program.

#include <stdio.h>
main () {
  int a;
  int b;
  int c;
  int d;
  static char buffer[]="1 2 3 4 ";
  int rc;
  rc=sscanf(buffer, "%d %n %d %n", &a, &b, &c, &d);
  fprintf(stdout, "rc: %d\na:  %d\nb:  %d\nc:  %d\nd:  %d\n", rc, a, b, c, d);
  exit(0);
}

When compiled with "cc test.c -O -o test", I get the following results.

rc: 2
a:  1
b:  2
c:  2
d:  4

This is what I expected .  When I compile with "gcc test.c -O -o test", I get
the following results.

rc: 1
a:  1
b:  0
c:  0
d:  0

It seems that %n is not processed correctly.  Is this a valid bug?  Note that
I am running gcc version 1.36 under SunOS Release 4.0.3 on a Sun-3/60M.  Any
help would be appreciated.

Thanks,
Geoff