Path: utzoo!utgpu!water!watmath!clyde!rutgers!mit-eddie!uw-beaver!cornell!batcomputer!pyramid!mips!mash From: mash@mips.UUCP (John Mashey) Newsgroups: comp.unix.wizards Subject: Re: Jerry Pournelle on UNIX (from BYTE) [really: whither UNIX] Message-ID: <1261@winchester.UUCP> Date: 9 Jan 88 21:32:20 GMT References: <11129@brl-adm.ARPA> <170@hdr.UUCP> <232@unicom.UUCP> Reply-To: mash@winchester.UUCP (John Mashey) Organization: MIPS Computer Systems, Sunnyvale, CA Lines: 89 In article <232@unicom.UUCP> greywolf@unicom.UUCP (When love and skill work together, expect a masterpiece.) writes: >In article <170@hdr.UUCP> eric@hdr.UUCP (Eric J. Johnson) writes: >> >>Wrong. a Unix rewrite is already in the works. This was mentioned >>by Bill Joy at the December Sun Users Group Convention in San Jose. >>The new 'enhanced' version of System V will be rewritten from scratch >>in C++. The slide he showed had all the major flavors of Unix being >>merged back into one product. >A rewrite in C--? Gads, that sounds like an absolute nightmare... >is C really going down and something else taking its place? The least >they could do if they were going to rewrite UNIX is to try and truly >integrate the best of both worlds (i.e. BSD & System V). (Each has its >own good points and drawbacks as I am sure most well-educated programmers >are aware. The problem with integrations is they end up integrating the >drawbacks and forgetting about the good points...) >Does C-- follow the ANSI standards? >Where is the UNIX world going?? Many of us have used the "Darwinian Selection" model of UNIX evolution for years; a) A new "standard" version of UNIX appears. [creation] b) Everybody takes it, adds extensions, changes to meet local needs. [mutation] c) After a while, people notice the chaos of having multiple versions that differ more than necessary, and there is a struggle to breed together the multiple versions, saving the good genes. [selection] d) Then, "a" happens again. This happened, in modest ways, for several years inside Bell Labs. It happened in a big way inside the Labs around 1977-1980, when upper management realized UNIX was critical to many projects, but varied randomly more than needed. There was a big, explicit effort to crunch together: stuff from research UNIX, USG, PWB, Columbus, etc. The BSD vs V thing is no different, although many people, especially those who aren't old-timers, seem to treat as a unique event. The natural state of UNIX, ever since it escaped from Lab 127 is; a) There are a bunch of features and interfaces that everyone agrees on. Those are "genes" that have been around a long time. b) There are a bunch of features where not everyone agrees on, but there are only a few ways that people do it, often for historical reasons. c) There are some things, especially those near the edge of the state of the art, or for applications that are more special-purpose, where no one agrees on anything, and it's mass chaos. Over time, items in b) turn into a), and c) turn into b), and new things appear in category c). The BSD vs V thing is in the b) turning into a) category right now. One has to ask if the proposed ATT+Sun V+BSD: a) will be the first such animal? [no: many other people already have heavily-merged versions right now. Look at, for example, HP/UX, among many.] Saying that the combined version will finally "end the BSD-V war" is one of the more amazing things I've seen: a lot of us thought that most people were ending the war by themselves anyway: take a look at the number of System-V based systems that have: sockets, BSD TCP/IP, BSD or other file systems, long pathnames, better signals, c-shell, etc, etc, OR the number of BSD-based systems that pass SVID, have shared memory, semaphores, etc, etc. b) can look a LOT different from what you'd expect? [how: AT&T is saying that people investing in SYS V code on 386s and 3Bs will be OK, and Sun is saying that SunOS user's will be OK also, and both are saying it will be POSIX-compatible and X/Open-compatible. Regardless of what's on the inside, it can't afford to break too many things on the outside.] Anyway, I don't see why integrations usually save the bad, and forget the good. I've been involved in several rounds of that in UNIX-land, and it usually hasn't been true. [If you think some yukky things got included, you should have seen some of the truly amazing things that got left out or cleaned up.] Also, there are enough reasonable BSD+V or V+BSD hybrids around to offer existence proofs. Now that AT&T is actually, finally, blessing the idea that it might be OK to include BSD features, maybe we can all be shipping our hybrids and not have to worry about whether or not it's heresy to have added sensible BSD features. -- -john mashey DISCLAIMER: UUCP: {ames,decwrl,prls,pyramid}!mips!mash OR mash@mips.com DDD: 408-991-0253 or 408-720-1700, x253 USPS: MIPS Computer Systems, 930 E. Arques, Sunnyvale, CA 94086 Brought to you by Super Global Mega Corp .com