From owner-freebsd-hackers  Sun Feb  2 17:50:39 1997
Return-Path: <owner-hackers>
Received: (from root@localhost)
          by freefall.freebsd.org (8.8.5/8.8.5) id RAA00413
          for hackers-outgoing; Sun, 2 Feb 1997 17:50:39 -0800 (PST)
Received: from dg-rtp.dg.com ([128.222.1.2])
          by freefall.freebsd.org (8.8.5/8.8.5) with SMTP id RAA00406
          for <hackers@FreeBSD.ORG>; Sun, 2 Feb 1997 17:50:37 -0800 (PST)
Received: by dg-rtp.dg.com (5.4R3.10/dg-rtp-v02)
	id AA22124; Sun, 2 Feb 1997 20:50:04 -0500
Received: from ponds by dg-rtp.dg.com.rtp.dg.com; Sun,  2 Feb 1997 20:50 EST
Received: from lakes.water.net (lakes [10.0.0.3]) by ponds.water.net (8.8.3/8.7.3) with ESMTP id UAA02990; Sun, 2 Feb 1997 20:17:54 -0500 (EST)
Received: (from rivers@localhost) by lakes.water.net (8.8.3/8.6.9) id UAA18377; Sun, 2 Feb 1997 20:22:10 -0500 (EST)
Date: Sun, 2 Feb 1997 20:22:10 -0500 (EST)
From: Thomas David Rivers <ponds!rivers@dg-rtp.dg.com>
Message-Id: <199702030122.UAA18377@lakes.water.net>
To: ponds!FreeBSD.ORG!hackers, ponds!NUXI.com!obrien
Subject: Re: conditionally including <sys/param.h>
Content-Type: text
Sender: owner-hackers@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk

> 
> Sorry to have to bring this back to light...
> I was under the impression that we had agreed that
> "#if (defined(__unix__) || defined(unix))" was the best way to
> conditionally include <sys/param.h>.  However, this email from Chuck,
> implies that this is a Bad Thing.

Unfortunately the symbols:
	unix
	FreeBSD
	NetBSD
	OpenBSD

violate the user's name space.  For example, I have programs
that define some of these on the command line.  If these were defined
elsewhere things would break.  [Of course, such things do break
on other hosts; e.g. SunOS 4.x defines "unix".]

I would suggest using:
	__unix__
	__FreeBSD__
	__NetBSD__
	__OpenBSD__

as identifiers beginning with "_" are reserved (per ANSI) for the
implementation.

	- Dave Rivers -


> 
> So to re-hash this, what does the list think is the best way to have an
> easy way to include <sys/param.h>?  Even thought I don't have all the
> *BSD folks in my grips like I did at USENIX, I'm still not above trying
> to get a new cpp symbol added (like __44bsd__ or something).
> 
> Chuck Robey writes:
> > > obrien      97/01/27 17:43:23
> > >   Modified:    share/doc/handbook  porting.sgml
> > >   Document the "#if (defined(__unix__) || defined(unix))" way of including
> > >   sys/param.h.  Change _HAVE_PARAM_H to "HAVE_SYS_PARAM_H" for those who
> > >   still like this method -- leading underscores are in the compiler/library
> > >   name space and the Ollivier says to follow GNU Autoconf anyway.
> > 
> > That test is completely bogus, cause it passes for all sysv systems,
> > which _don't_ have sys/param.h.  It's a completely useless test, and
> > someone justified it by saying that "most systems have a sys/param.h
> > anyhow".
> > 
> > What (in my own opinion) should have been done is to detect:
> > 
> > #ifdef FreeBSD || NetBSD || OpenBSD
> > 
> > that would have worked for those variants.  Would be able to find out the
> > one for BSDi systems, then we'd have a great deal of the important ones.
> > That's how X11R6 does it.
> 
> -- 
> -- David	(obrien@NUXI.com  -or-  obrien@FreeBSD.org)
>