tut@ucbopal.CC.Berkeley.ARPA (10/01/84)
According to an advertisement for SCO in the latest "Unix Review", Xenix constitutes almost 80% of installed Unix systems on micros. I didn't believe it at first. But in a recent issue of "The Yates Perspective" (a marketing newsletter), there was a pie chart of installed Unix systems, broken down as follows: Xenix 77% Version 7 20% System III 3% (System V must be 0%) The funny thing was that another pie chart showed that 23% of the vendors sell System III, thus chasing 3% of the market! Frankly, this shocked me. Up till now I've believed AT&T's propaganda that System V will become the standard Unix-- an indication, I thought, that the market can't make informed technical decisions. Maybe the market is smarter than I thought. What does System V have that Version 7 doesn't have, besides termcap and vi (which have been in Xenix for a long time)? Shared memory? (big deal). An incompatible terminal driver? An incompatible init? A few incompatible library routines? The cut and paste programs? Remember that Xenix is Version 7 based, with Berkeley enhancements, and System III compatibility. Thus, 97% of Unix micros are running Version 7. The IBM PC/AT has been announced with Xenix, and although I've heard rumors that Interactive's (System III based) PC/IX will be available for it, this hasn't been stated in any IBM advertising. If I were writing piece of commercial software, I would make damn sure it ran on the PC/AT under Xenix, and I would strive to eliminate all System V dependencies in the code! Bill Tuthill (as a private citizen)
jrb@wdl1.UUCP (jrb ) (10/02/84)
I suspect that most 'System III' ports are in fact V7 ports with the System III stuff tacked on round the edges. John R Blaker UUCP: ...!fortune!wdl1!jrb ARPA: jrb@FORD-WDL1 and blaker@FORD-WDL2
marcum@rhino.UUCP (Alan M. Marcum) (10/02/84)
An interesting question regarding the statistics Bill Tuthill quoted
[vis. Xenix 77%
Version 7 20%
System III 3% (System V must be 0%)]
is, to what do these statistics pertain? Is this a measure of
the development genesis of the systems, of the licensing path
taken by the various manufacturers, or what?
One would expect that several thousand (yes, THOUSAND) Un*x
installations would show up SOMEWHERE in the above figures, yes?
And yet, Fortune Systems ("For:Pro") is nowhere listed. UNLESS,
of course, we talk about the licensing agreements, in which case
For:Pro is included in the Xenix figure.
Note that For:Pro is NOT a Xenix re-port. Neither is For:Pro
Xenix based. For financial reasons, though, Fortune's LICENSE is
based on the Xenix license.
As far as the technical port is concerned, For:Pro is V7 based,
with many of the 4.1 commands included. Fortune has made many
kernel modifications, including several to the scheduler, new I/O
drivers, and a mechanism for automatically determining the device
controllers attached to the system and accessing the appropriate
device drivers dynamically.
(Note that these statements are NOT official Fortune positions,
merely observations based on my experiences.)
--
Alan M. Marcum Fortune Systems, Redwood City, California
...!{ihnp4, ucbvax!amd, hpda, sri-unix, harpo}!fortune!rhino!marcumbass@dmsd.UUCP (John Bass) (10/03/84)
A couple of things to adjust your numbers of the installed base and who runs what by: I'm not sure where SCO got their numbers but it is likely the The AT&T number of licenses sold. Two things make the installed base for XENIX look larger than it really is. First Fortune 32:16's have a UNIX license thru Microsoft's XENIX ... the Fortune doesn't run XENIX as distributed by Microsoft. Fortune did it's own M68k port which is a mixture of V7 & 4.1. The Microsoft guys gave Fortune a good deal on the price ... and Fortunes volume pushed XENIX quickly down the AT&T curve. Secondly from what I can tell all tandy model 16 hard disk expansions come with a xenix license. It is clear that some number of these are not really running UNIX ... but are really being used as hard drives for CPM using the Mod 16 as a mod 12 ... or were simply installed on a mod 12. Fortune and Tandy represent the large share of XENIX's ... I have forgotten the numbers. Any one else know the breakdown? Secondly, it takes 12-18 months to bring a new kernel release to high volume production release ... this is a non-trival effort tomake sure that when you make 50,000 copies that you don't have to make 50,000 updates the next month to fix a major bug. Microsoft XENIX OEMS have been shipping some version of 2.x .. which is V7 based. Microsoft is just now releasing versions of 3.x which is system III/V ... it will take 6-12 months for many of the installed base to decide to upgrade. After that time the distribution of who runs what will by greatly biased toward System V. Nearly every major vendor is releasing a System V upgrade or new product in the next year ... it truly is on the verge of being the DEFACTO standard by the second half of '85. By this time next year I expect that the installed base for V7 + 4.x will be something much less than 10% -- the balance will be System 3/5 Any comments? John
ron@BRL-TGR.ARPA (10/04/84)
From: Ron Natalie <ron@BRL-TGR.ARPA> The reason I suspect that everyone is running Xenix rather than either of the two stock AT&T releases (by the way, I don't trust anything Yates says) is that they are not for micros. They are written for PDP-11's and VAX's. If you were a micro manufacturer, would you go out and do your own unix port of the AT&T code, if you could get one for your processor already from Microsoft. Microsoft is in the business of making micro UNIX ports, AT&T isn't (or if they are, they aren't doing anything yet). I really think that the quality of the different versions was not a major factor. Xenix is a nice product however. I got to stick my nose into Xenix III for the 80286 and was fairly impressed. -Ron