*BSD News Article 40943


Return to BSD News archive

Newsgroups: comp.os.386bsd.misc
Path: sserve!newshost.anu.edu.au!harbinger.cc.monash.edu.au!newshost.marcam.com!news.mathworks.com!noc.near.net!monk.proteon.com!jfw
From: jfw@proteon.com (John Woods)
Subject: Re: To Merge or Not to Merge *BSD. What does it really mean?
Message-ID: <D2KKxB.Euq@proteon.com>
Sender: news@proteon.com
Nntp-Posting-Host: kerplop.proteon.com
Organization: Proteon, Inc., Westborough, Ma.
References: <3enogm$5l7@fw.novatel.ca> <3f1h83$hgl@newshost.lanl.gov> <3f253c$es0@agate.berkeley.edu> <3f9hst$q8n@idiom.com> <3fbgok$s0i@ivory.lm.com>
Date: Tue, 17 Jan 1995 21:55:59 GMT
Lines: 28

peterb@telerama.lm.com (Peter Berger) writes:
>In article <3f9hst$q8n@idiom.com>, David Muir Sharnoff <muir@idiom.com> wrote:
>>The FreeBSD camp should consider it a bug whenever a NetBSD program 
>>fails to compile or its binary to run.
>>The NetBSD camp should consider it a bug whenever a FreeBSD program 
>>fails to compile or its binary to run.
>This is the single most intelligent thing that has been said in the
>entire discussion.

The problem with the above statement is that both NetBSD and FreeBSD camps
plan to do some of their own innovations, which will probably eventually
lead to enough divergence that there will exist NetBSD programs that it
is impractical for FreeBSD to support and vice versa.

What makes more sense is for both camps to agree on a common Application Binary
Interface, so that it will be practical to run common applications on either
OS platform; kernel system calls that aren't developed jointly (or at least
quickly adopted by the other platform) should be kept out of the common ABI,
and some effort should be made to ensure that applications don't use such
extensions casually (hence, try not to invent a "better" open() system call).
It's unfortunate that the shared library implementations are different, since
that's a cool way to achieve a clean ABI (IMHO, anyay).

Actually, several ABIs are required, since neither port is limited to the 386.

In terms of compilation, both camps have (I believe) the same committment to
maintain and improve the POSIX compliance of 4.4BSD, so for the most part
that should take care of itself.