*BSD News Article 6303


Return to BSD News archive

Xref: sserve comp.unix.sysv386:24584 comp.unix.sys5.r4:188 comp.unix.sys5.r3:35 comp.unix.pc-clone.32bit:287 comp.unix.bsd:6351 comp.os.linux:12223 comp.os.mach:2272 comp.windows.x:46103
Newsgroups: comp.unix.sysv386,comp.unix.sys5.r4,comp.unix.sys5.r3,comp.unix.pc-clone.32bit,comp.unix.bsd,comp.os.linux,comp.os.mach,comp.windows.x
Path: sserve!manuel.anu.edu.au!munnari.oz.au!spool.mu.edu!umn.edu!paperboy.micro.umn.edu!magnus.acs.ohio-state.edu!zaphod.mps.ohio-state.edu!usc!wupost!uunet!portal!kdenning
From: kdenning@portal.hq.videocart.com (Karl Denninger)
Subject: Re: Announcing the availability of XFree86 1.1
Message-ID: <Bvv4Ay.9GL@portal.hq.videocart.com>
Summary: More on this and unbundling
Organization: VideOcart Inc.
References: <1992Oct3.215041.17541@cbnewsj.cb.att.com> <1992Oct7.194931.16296@crd.ge.com> <Bvux69.Az2@chinet.chi.il.us>
Date: Fri, 9 Oct 1992 16:25:45 GMT
Lines: 34

In article <Bvux69.Az2@chinet.chi.il.us> randy@chinet.chi.il.us (Randy Suess) writes:
>In article <1992Oct7.194931.16296@crd.ge.com> davidsen@crd.ge.com (bill davidsen) writes:
>>| This, by the way, is a perfect example of why SVR4 is better than SCO.
>>| We have a single binary kit, ~20MB compressed, that works on SVR4 from 7 
>>| different vendors.  
>>This may come as a shock, but V.3 worked that way, too, and Xenix/386
>>binaries will run on virtually anything. 
>
>	Bill, I have an idea that this refers to the fact that most
>	(if not all) r4s are pretty standard, specially in the networking
>	code.  Believe me, a networking program like X is NOT compatible
>	with all svr3's.  Back then, everyone did their network interface
>	different than everyone else.

Let me poke at that a bit.

The REASON that this is true in SVR4 is that there is a >bundled< IP stack
and assorted related items in the base OS.  Thus, the standardization which
makes possible the distribution of binary objects using these facilities is
a result of the bundled approach.

SVR3.x didn't work this way because TCP/IP was an add-on.  As a result every
maker did it differently, and the libraries from one would not work with the
other.  Thus, there was no standardization.

This, IMHO, is one of the best arguments for bundling a known set of
functionality in a release.  By doing this you nearly guarantee
compatibility across vendors.  This is a powerful argument for SVR4 and
related items...

-- 
Karl Denninger 		Inet:  kdenning@hq.videocart.com
VideOcart Inc.		Voice: (312) 987-5022