mike%brl-vgr@sri-unix.UUCP (11/11/83)
From: Mike Muuss <mike@brl-vgr> Dave - When the "standard" version of UNIX from AT&T begins to stay "state of the art", and offers decent network support, virtual memory, fast disk I/O, etc, etc -- THEN, and only then, will I embrace the "standard". Alas, (I had best duck the incomming missiles), the word "standard" brings to mind things like COBOL, and ADA. Useful items, with an important place in the world -- as long as it is far away from me! The only good standard that comes to mind in recent times is TCP/IP. (But thats another story). Now, in this specific instance (AT&T -vs- Berkeley), things are not so cut-and-dried; Berkeley has the hottest kernel, but Bell has the best polish. Improved utilities, better manuals, etc, etc. And, in the finest tradition of hacking ...er, ahem... computer science, BRL is doing something about it! Thanks to the tireless efforts of Mr. Douglas Gwyn here at BRL, there now exists a full System V environment operating and co-existing with the 4.2 BSD environment (well, 4.1c, until our bloody 4.2 tape gets here). Merely by diddling your $PATH variable, you can chose which world you wish to live in. (Me? I have two CSH aliases "sysv" and "std", to let me switch between them). So, what should everybody else do? For sites with "smarts", buy a System V license, bring up 4.2 BSD, add on the BRL System_V-on-4.2, and enjoy the best of both worlds. The rest of you folks get to sit and suffer! Welcome to the wonderful world of VENDOR SUPPORT. Yes, you too can discover that vendor support for UNIX can be as bad as vendor support for VMS, or for NOS, or for OS/VS2, or for ..... Go lean on your vendor. See where it gets you. A good example of vendor support might be the integration of TCP/IP into vendor operating systems. TCP for VMS? No. TCP for AT&T UNIX? No. TCP for VM? No. Not from the vendors. Funny thing is, TCP is even a standard. And DoD buys ~7 % of all the computers in America. Army has even been paying vendors to develop TCP, for inclusion in their "standard" operating system. Generous, huh? Well, it's over a year late. "Standard vendor support". (Yuck -- the 3 words I hate most, in 1 sentence! I can't stand it). Being big doesn't help. DEC doesn't care. AT&T doesn't care. IBM doesn't care. At least Berkeley (with a few $M from DARPA) had the guts to go out and build a system that's pretty close to "state of the art" for this generation of operating system. For that we should thank Bill Joy and Sam Leffler and the rest of the gang at Berkeley. Tired of "Standard" and Looking for "Good", I Remain, Truely Yours, Zzzzzz, -Mike Muuss U.S. Army BRL.
obrien%rand-unix@sri-unix.UUCP (11/11/83)
world in the form of a major release is basically "nice guy" work, because increasing the reputation of Berkeley and the pool of outside users makes research easier. For an example of what 4.2 would be like without the DARPA funding, you have only to look at 2.9BSD...except it would be worse, since much of 2.9 comes from 4.2. Those folks who think that 4.2 is nice but that Berkeley should get their act together and behave like a proper vendor are not cognizant of the fact that they are merely the recipients of the fallout of government research, NOT of a development campaign mounted by folks out for commercial bucks. Berkeley's ethical responsibility lies solely with DARPA and its own grad students.
wert.rice%rand-relay@sri-unix.UUCP (11/12/83)
From: Scott Comer <wert.rice@rand-relay> I would like point out that just about anything that comes out of Berkeley is hardly "state of the art". As a computer scientist, "state of the art" means to me that the system is progressive in its functionality, robust in its implemetation, clear in its documentation, etc. 4.* is none of these things. There is no modular design. There is no design at all. Robust? Don't make me laugh. And the documentation often does, when I am not crying... The only things that Berkeley has going for it are that it is cheap, you get the source so that you can make it work, and there is a lot of necessary software that only runs on it... wert
jab@uokvax.UUCP (11/16/83)
#R:sri-arpa:-1355900:uokvax:6200003:000:748 uokvax!jab Nov 14 16:07:00 1983 I can't stand the comments to the effect of "Berkeley should be supporting their spiffy kernel or shouldn't distribute it". First, O'Brien's comments that the ethical responsibilities at UCB lie with DARPA and Berkeley graduate students is on the nose. No university in its right mind would ever try to become a software warehouse (and "no, Berkeley isn't one") --- a university's responsibilities are those of research and education (one implies the other, but I forget in which direction the implication points). Second, the AT&T/Berkeley debates are at a standoff. I've heard a lot of pro-Berkeley debates, and some pro-AT&T debates. Haven't you noticed that both ends of the spectrum are SLOWLY moving back together? Jeff Bowles Lisle, IL