X-Authentication-Warning:	delorie.com: mailnull set sender to djgpp-workers-bounces using -f
From:	sandmann AT clio DOT rice DOT edu (Charles Sandmann)
Message-Id:	<10202071825.AA14352@clio.rice.edu>
Subject:	Re: conflicting types for bzero (gcc303)
To:	djgpp-workers AT delorie DOT com
Date:	Thu, 7 Feb 2002 12:25:19 -0600 (CST)
Cc:	eliz AT is DOT elta DOT co DOT il (Eli Zaretskii), ST001906 AT HRZ1 DOT HRZ DOT TU-Darmstadt DOT De,
	pavenis AT lanet DOT lv
In-Reply-To:	<1013078746.9713.25.camel@bender.falconsoft.be> from "Tim Van Holder" at Feb 07, 2002 11:45:45 AM
X-Mailer:	ELM [version 2.5 PL2]
Mime-Version:	1.0
Reply-To:	djgpp-workers AT delorie DOT com
Errors-To:	nobody AT delorie DOT com
X-Mailing-List:	djgpp-workers AT delorie DOT com
X-Unsubscribes-To:	listserv AT delorie DOT com

Okay, bzero only generates warnings (no errors) if:

 1) GCC 3.x, AND
 2) Only if -I /djgpp/include is used
 3) and -fno-builtin not specified
 
Do we do anything to the refresh at all?  (It's fixed in CVS and
has been for quite some time).  If we wanted to remove warning:

A quick test of Andris's suggestion (in a GNUC 3 ifdef) :
#define bzero(s,n) __builtin_bzero(s,n)

Also seemed to work fine (even with -fno-builtin)

Note that if we make this change we always use the builtin with
GCC 3.x; if we don't we always seem to generate code for our libc 
bzero (backward compatible but gives occasional warning).

So a simple string.h ifdef for __GNUC__ >=3 to use the above
instead of our prototype would fix the warning - and upgrade
you to posix behavior using GCC 3.x.  While this would still
break anyone trying to use a return argument, they probably need
to make other code changes for GCC 3.x anyway, or could use 2.x.

This then asks - should we be using more builtins in the headers
for CVS?

There was also a discussion about bcopy - but this doesn't seem
to generate any warnings with GCC 3.03 that I saw - so I don't
see a reason to mess with it in any case.  (Just confusion with
the bzero issue, or something else?)

- Raw text -