From owner-freebsd-ports  Tue Jan  7 23:50:01 1997
Return-Path: <owner-ports>
Received: (from root@localhost)
          by freefall.freebsd.org (8.8.4/8.8.4) id XAA19677
          for ports-outgoing; Tue, 7 Jan 1997 23:50:01 -0800 (PST)
Received: from dfw-ix10.ix.netcom.com (dfw-ix10.ix.netcom.com [206.214.98.10])
          by freefall.freebsd.org (8.8.4/8.8.4) with SMTP id XAA19646
          for <ports@freebsd.org>; Tue, 7 Jan 1997 23:49:58 -0800 (PST)
Received: from silvia.HIP.Berkeley.EDU (wck-ca21-12.ix.netcom.com [207.94.231.108]) by dfw-ix10.ix.netcom.com (8.6.13/8.6.12) with ESMTP id XAA21614; Tue, 7 Jan 1997 23:47:27 -0800
Received: (from asami@localhost) by silvia.HIP.Berkeley.EDU (8.8.4/8.6.9) id XAA11117; Tue, 7 Jan 1997 23:47:19 -0800 (PST)
Date: Tue, 7 Jan 1997 23:47:19 -0800 (PST)
Message-Id: <199701080747.XAA11117@silvia.HIP.Berkeley.EDU>
To: m230761@ingenieria.ingsala.unal.edu.co
CC: ache@nagual.ru, imp@village.org, ports@freebsd.org
In-reply-to: <Pine.SOL.3.91.970107181336.1376A-100000@ingenieria.ingsala.unal.edu.co> (message from Pedro Giffuni on Tue, 7 Jan 1997 18:34:21 -0500 (EST))
Subject: Re: Niklas Hallqvist: archivers/hpack.non-usa.only
From: asami@cs.berkeley.edu (Satoshi Asami)
Sender: owner-ports@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

 * I disagree with this (proposed) way of patching: as clearly stated on "the 
 * porting guidelines", there is a common way to identify BSD4.4, if there are 
 * differences, then we should use the specific *BSD flag, but it doesn't seem 
 * logical to treat everyone as a special case of FreeBSD. Changes in the 
 * diffs will have to be made anyway, but the general idea is that patches 
 * should disappear after we feedback the diffs to the author of the 
 * program. I wouldn't like to define FreeBSD, OpenBSD, and NetBSD because 
 * the port was submitted by someone in another BSD camp.

I agree with this in principle.  If it can be differentiated by the
BSD macro, it should be used instead of __FreeBSD__ and/or
__OpenBSD__.

By the way, has anyone have a list of changes that can be
distinguished by the BSD macro?  I haven't been around the sources too
long, and I'm sure there will be others there that will appreciate it
(we can even put it in the handbook!)....

 * > Maybe just add -D__FreeBSD__ to OpenBSD CFLAGS in bsd.ports.mk instead and
 * > handle differences in this way:
 * > 
 * > #ifdef __FreeBSD__
 * > #ifdef __OpenBSD__
 * > OpenBSD case
 * > #else
 * > FreeBSD case
 * > #endif
 * > #endif
 * > 
 * > As I think OpenBSD will match FreeBSD in 99% cases.

I don't think OpenBSD people would like that. ;)

And about the patches:

 * > > ! + #if !defined (__FreeBSD__) && !defined(__OpenBSD__)
 * > >     	gets( password );
 * > > ! + #else   /* gets depricated in FreeBSD && OpenBSD */
 * > >   + 	fgets( password, sizeof(password), stdin);
 * > >   + 	len = strlen(password);
 * > >   + 	if (password[len - 1] == '\n')

gets() is unsafe on all operating systems, I don't see why we should
condition this on *BSD....

 * > > ! + #if defined (__FreeBSD__) || defined (__OpenBSD__)
 * > >   + found:
 * > >   + #endif

No idea on this one.

 * > >     #if defined( BSD386 ) || defined( CONVEX ) || defined( HPUX ) || \
 * > >     	defined( IRIX ) || defined( LINUX ) || defined( POSIX ) || \
 * > > ! ! 	defined( SVR4 ) || defined( UTS4 ) || defined( __FreeBSD__ ) || \
 * > > ! ! 	defined( __OpenBSD__ )
 * > >     #include <termios.h>

Argl.  Shouldn't BSD/OS, FreeBSD and OpenBSD all use BSD, or are there
systems with non-functional termios.h?

 * > >     #if defined( __UNIX__ ) && !( defined( BSD386 ) || defined( CONVEX ) || \
 * > > ! ! 							  defined( NEXT ) || defined( ULTRIX_OLD ) || defined( __FreeBSD__ )) || defined( __OpenBSD__ )
 * > >       #include <malloc.h>				/* Needed for mem.functions on some systems */

You've got the parentheses wrong here, or it will include <malloc.h>
in OpenBSD too (or is it supposed to do that?).

 * > >     #if defined( AIX386 ) || ( defined( __AMIGA__ ) && !defined( LATTICE ) ) || \
 * > >     	defined( __ARC__ ) || defined( __ATARI__ ) || defined( __MSDOS16__ ) || \
 * > > ! ! 	defined( __MSDOS32__ ) || defined( __FreeBSD__ ) || defined( __OpenBSD__ )
 * > >       #include <stdlib.h>				/* Needed for mem.functions */

This comment contradicts the OpenBSD case above, so I guess it should
have been inside the parens.

Well, I'm going to quit here.  But boy, this code is ugly.

Satoshi