Sender: rich AT phekda DOT freeserve DOT co DOT uk
Message-ID: <3E2D52C7.70E1DDEB@phekda.freeserve.co.uk>
Date: Tue, 21 Jan 2003 14:01:43 +0000
From: Richard Dawe <rich AT phekda DOT freeserve DOT co DOT uk>
X-Mailer: Mozilla 4.77 [en] (X11; U; Linux 2.2.23 i586)
X-Accept-Language: de,fr
MIME-Version: 1.0
To: djgpp-workers AT delorie DOT com
Subject: Re: size_t and ssize_t
References: <200106192104 DOT XAA18859 AT father DOT ludd DOT luth DOT se>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Reply-To: djgpp-workers AT delorie DOT com

Hello.

Going back a bit here: the message quoted below is dated Tue, 19 Jun 2001
23:04:41 +0200 (MET DST) on my system.

Martin Str|mberg wrote:
> 
> According to DJ Delorie:
> >
> > > Any particular reason for not having "#define __DJ_ssize_t
> > > typedef long int ssize_t"? Or even "long signed int ssize_t"?
> >
> > There is no reason for or against any change.  Nobody is supposed to
> > know what those types are, so it shouldn't matter what they are.  Is
> > there a specific reason for the change?
> 
> I thought ssize_t was the signed couterpart of size_t. If so, they
> should be similarly defined. Like they are now, when I need to cast
> from size_t I currently use "(int)" because there isn't any similarity
> between size_t and ssize_t. If ssize_t == signed size_t, then using
> "(ssize_t)" would be more appropriate.
> 
> Look at my signed/unsigned mail to follow soon.

Well, we still seem to have ssize_t defined as an int and size_t defined as a
long. I'm currently coding up support for the 'z' conversion qualifier for
printf. This specifies that the following integer is of size_t or some
equivalent sized thing. This is the kind of code that should work IMHO:

    #include <stdio.h>
    #include <stdlib.h>
    #include <sys/types.h>

    int
    main (void)
    {
      ssize_t foo = 5;

      printf("%zd\n", foo);
      return(EXIT_SUCCESS);
    }

When I compile it, I get this warning:

gcc @../../../gcc.opt -I. -I- -I../../../../include  -c foo.c
cc1.exe: warnings being treated as errors
foo.c: In function `main':
foo.c:17: warning: signed size_t format, ssize_t arg (arg 2)
make.exe: *** [foo.o] Error 1

If I change the program to define ssize_t like this:

    typedef long int ssize_t;

instead of including <sys/types.h>, then I get no warning.   

I haven't tried changing <sys/djtypes.h> to define ssize_t as a long. Has
anyone tried it?

Thanks, bye, Rich =]

-- 
Richard Dawe [ http://www.phekda.freeserve.co.uk/richdawe/ ]