RE: Traditional numbers

[Bernard Dautrevaux <Dautrevaux at microprocess dot com>]

> -----Original Message-----
> From: James Grosbach [mailto:james.grosbach@microchip.com]
> Sent: Monday, July 30, 2001 5:09 PM
> To: gcc@gcc.gnu.org
> Subject: Re: Traditional numbers
> 
> 
> On Fri, Jul 27, 2001 at 06:35:48PM +0100, Neil Booth wrote:
> > int main()
> > {
> > #if 0xffffffff < 0
> >     puts("cpp:negative");
> > #else
> >     puts("cpp:positive");
> > #endif
> >     puts(0xffffffff < 0 ? "c:negative" : "c:positive");
> >     return 0;
> > }
> > 	bash$ gcc3 -traditional a.c && ./a.out
> > cpp:positive
> > c:negative
> > 
> > Does the list agree that the above behaviour is a bug?
> 
> Yes. According to my understanding of the standard, the compiler
> is in error. The integer constant should be of the first type
> from the following list into which the value will fit: int,
> unsigned int, long int, unsigned long int. The value 0xffffffff
> will not fit in a signed 32-bit type, so the type of the
> constant is unsigned int, and the comparison should yield
> false.

I would agree with you, except that the "-traditional" flag here is bending
gcc towards "traditional" C, as opposed to ANSI C. I agree that in ANSI C
0xffffffff is positive; however in traditional C is may be (and often is)
negative :-)

HTH

	Bernard.

PS: in this case the problem is probably in the preprocessor, which with
"-traditional" should keep the same conventions as the compiler for the sake
of consistency, although K&R was not precise enough to be able to qualify
this of a BUG :-)

--------------------------------------------
Bernard Dautrevaux
Microprocess Ingenierie
97 bis, rue de Colombes
92400 COURBEVOIE
FRANCE
Tel:	+33 (0) 1 47 68 80 80
Fax:	+33 (0) 1 47 88 97 85
e-mail:	dautrevaux@microprocess.com
		b.dautrevaux@usa.net
--------------------------------------------