From mboxrd@z Thu Jan  1 00:00:00 1970
From: Wilco van Hoogstraeten <wilco@trimedia.com>
To: nobody@gcc.gnu.org
Cc: gcc-prs@gcc.gnu.org
Subject: Re: c/4302: wrong code generated
Date: Wed, 12 Sep 2001 09:06:00 -0000
Message-id: <20010912160601.20533.qmail@sourceware.cygnus.com>
X-SW-Source: 2001-09/msg00248.html
List-Id: <gcc-prs.sourceware.org>

The following reply was made to PR c/4302; it has been noted by GNATS.

From: Wilco van Hoogstraeten <wilco@trimedia.com>
To: Andreas Jaeger <aj@suse.de>
Cc: wilco@trimedia.com, gcc-gnats@gcc.gnu.org
Subject: Re: c/4302: wrong code generated
Date: Wed, 12 Sep 2001 08:59:01 -0700

 Andreas Jaeger wrote:
 > wilco@trimedia.com writes:
 > 
 [...]
 > > >Description:
 > > Wrong code generated:
 > >
 > > marilyn % cat tmp.i
 > > extern int printf( const char* args, ... );
 > >
 > > int main( )
 > > {
 > >   long long X, Y;
 > >
 > >   Y = -3000000000LL > (X = 0);
 > 
 > There's no bug here: X is not initialized at this point.  Therefore
 > the value of (X=0) is undefined,
 > 
 Excuse me???? The value of (X=0) is defined and is 0.
 
 Anyway, if you do not believe in C semantics, you could have tried
 to initialise X before rejecting the bug:
 
 extern int printf( const char* args, ... );
 
 % cat tmp.i
 int main( )
 {
    long long X = 0, Y;
 
    Y = -3000000000LL > (X = 0);
 
    printf( "%s (%lld)\n", Y != 0 ? "FAIL" : "OK", Y );
 
    return 0;
 }
 % gcc tmp.i
 % a.out
 FAIL (1)
 %
 
 Wilco.