X-Authentication-Warning: delorie.com: mail set sender to djgpp-workers-bounces using -f
Date: Mon, 25 Feb 2008 09:54:02 -0500
Message-Id: <200802251454.m1PEs25a005733@envy.delorie.com>
From: DJ Delorie <dj AT delorie DOT com>
To: djgpp-workers AT delorie DOT com
In-reply-to: <200802251436.08936.juan.guerrero@gmx.de> (message from Juan
	Manuel Guerrero on Mon, 25 Feb 2008 14:36:08 +0100)
Subject: Re: Documentation about header structure
References:  <200802251436 DOT 08936 DOT juan DOT guerrero AT gmx DOT de>
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
Precedence: bulk


> I have found that they use functions declared in argz.h and
> asprintf.  I am aware that these are neither ANSI nor posix
> functions and that the functions declared in argz.h may be provided
> by gnulib.  Neitherless this rises the question for interested
> people how this can be implemented for djgpp and thus what the
> structure of a header should looks like?

> Is there some definitive documentation about how a header that may
> declare ansi and/or posix and/or BSD functions should looks like?

Start by copying the _ansi template.  Fill in the blanks in the
#-lines.  The template defines three sections: ansi prototypes, posix
prototypes, and everything else.  In your case, only the "everything
else" section has content.

The whole issue with stubs and __ff functions happens when an ansi or
posix function needs a non-ansi-non-posix function to do its work
inside libc.  For those, we need to restrict ourselves to
namespace-clean names to preserve the ansi-clean or posix-clean status
when the depending ansi/posix functions are used.  For those cases, we
use the __ff stubs.

For plain old "extra stuff" we just add it in as-is.