Unix bugs vs. VMS bugs
Mark Crispin
This sort of issue comes up whenever people get the
impression that there are any absolutes. Just about any system
can benefit from having an on-site wizard, even if the operating
system is manufacturer-supported (e.g. VMS, TOPS-20, VM/370).
While the cost of ownership of a wizard is non-trivial (yes, they
do "spend lots of time on the phone, going to conferences,
reading nets like this, and hacking"), consider the alternative.
You are either stuck with the product as it comes from the
manufacturer or you find yourself forced to rent a wizard -- that
is, you must hire a consultant.
Now I have nothing against consultants! I'm a full-time
rental wizard (tr: independent consultant) and I find the
business quite lucrative. I hope that attitudes such as Jon
Forrest's continue -- customers with that attitude comprise most
of my business.
The "people problem" with Unix is not the wizards, but
rather the groupies. I define a "Unix groupie" as any individual
who (1) considers Unix in its present state to be software
perfection, (2) refuses to believe that other operating systems
have features too, (3) makes noises of disgust whenever some
other operating system is mentioned, (4) makes noises of disgust
whenever some programming language other than C is mentioned.
It's reminiscent of the APL\360 groupies of 15 years ago.
Unix does have a software maturity problem. I for one would
love to see a standard Unix. It unnerves me when I must relearn
"how to do X" just because I'm using somebody else's Unix system.
Many of these incompatibilities seem to be completely gratuitous.
Also, Unix lacks some very basic facilities which are only now
starting to appear: process-to-process memory mapping (for both
read and write), process-to-file memory mapping, file interlocks,
long file names, user-friendly command interfaces (sh, csh, ksh,
etc. are many things, but user-friendly is not one of them), etc.
I wish that these things would all appear in all places in the
same way, but I fear that in just about every minor version of
Unix it'll be completely different.
Unix is clearly not for the fainthearted. If you really
don't care all that much what the operating system does for you
-- e.g. all you want is a FORTRAN engine -- then Unix may not be
your answer. You can use a "throwaway" operating system such as
VMS. If you actually start USING some special feature of your
operating system, you may start caring about what happens when
you have to change computer vendors.
Finally, I cannot let the comment about "Unix being better
than any other operating system (except VMS)" go by unchallenged.
I can't see how anybody can possibly make such grand claims about
VMS. It's the manufacturer-supplied operating system for a
superminicomputer which is now (with the 8600) selling at (high)
mainframe prices. It's an upgrade from an earlier minicomputer
operating system from that manufacturer, but still some years(!)
away from achieving the level of functionality of other operating
systems from that manufacturer's other product lines! It's still
a dinosaur.
Mark Crispin
M...@SU-SCORE.ARPA
-------
ti...@cmcl2.uucp
>Oh, you say you have an unsupported copy of UNIX? Whose fault is that?
Do you see many people supporting UNIX with virtual address spaces
and IP/TCP around?
The fact that there are a few who may be able to offer this in
the near term future (one, two, three years tops) is heartening.
But nothing to get too carried away about.
Doug Gwyn <gwyn>
> and IP/TCP around?
Yes.
ti...@cmcl2.uucp
>> Do you see many people supporting UNIX with virtual address spaces
>> and IP/TCP around?
>Yes.
Neat. We could use someone who will provide Unix support, including IP/TCP,
full virtual memory, the usual bunch of programs written to run on 4.2 BSD
on about 4 or 5 VAXen for prices comperable or lower than DEC's in the Basic
and "Self-Maintenance" categories of Software Support for the comperable
service levels. Of course they should be a major company committed to the field
for at least the next five to ten years.
What are the names of all these support providers?
Ron Natalie
Memory support NOW.
-Ron
Ron Natalie
by their system today you get 4.1c. 4.2 will be available after the new
year. We've ordered their computers primarily since their performance beats
the hell out of anything DEC offers and the price is lower.
-Ron
Spencer W. Thomas
Gould 97/32. I was mucho impressed with the speed. I also got to run
on a pre-release version of their 4.2, and it seems to work, anyway.
Didn't try the virtual memory version, though.
Unfortunately for them, the new DEC 8600 comes in at about the same
performance, and about the same price. We have decided that we probably
won't buy an "off brand" unless it offers significantly better
price/performance than a VAX. This is to reduce the need for producing
multiple versions of software - currently, we can compile once and
distribute binaries (except the Suns, and they are always behind the
rest of the world here). To offset this hassle, a competitor must
provide some definite advantage. 6 months ago, Gould had the advantage
- you couldn't get a VAX with that performance. Now, it isn't true.
Sigh.
=Spencer
Ron Natalie
Certainly beats the DEC entry.
-Ron
Ted Nolan
>
>Neat. We could use someone who will provide Unix support, including IP/TCP,
>full virtual memory, the usual bunch of programs written to run on 4.2 BSD
>on about 4 or 5 VAXen for prices comperable or lower than DEC's in the Basic
>and "Self-Maintenance" categories of Software Support for the comperable
>service levels. Of course they should be a major company committed to the field
>for at least the next five to ten years.
>
>What are the names of all these support providers?
Isn't this what DEC is doing with Ultrix? (I would certainly expect their
prices to be comparable with DEC's :-) )
Ted Nolan ..usceast!ted
--
-------------------------------------------------------------------------------
Ted Nolan ...decvax!mcnc!ncsu!ncrcae!usceast!ted
6536 Brookside Circle ...akgua!usceast!ted
Columbia, SC 29206
("Deep space is my dwelling place, the stars my destination")
-------------------------------------------------------------------------------
Mike Muuss
I have benchmarked the Gould 8780 at 9.9 x a VAX 780 doing
double precision & heavy pointer math.
All claims I have heard for the DEC 8600 rate it as 3 - 4 x a VAX 780.
Has this changed?
-Mike
John Gilmore
> and IP/TCP around?
Yes.
Frederick M. Avolio
>Sorry, but the Gould PN9000 is about 10 times a 780 for about $400K.
>Certainly beats the DEC entry.
>
>-Ron
Now, do you believe that stuff? I didn't say it wasn't true... I asked if
you believed it! I won't, aside from that, comment on Ron's observation as
it pains me to do so. (By the way, Ultrix-32 is 4.2BSD based...)
But yes, such support is available through the companies mentioned (and
others I am sure). Ultrix-32, for example, is a *supported* Digital S/W
product. That means manuals, training, bug fixes, newsletters, updates, an
800 number hot line for customer help (already in use), a large group of
UNIX experts, and software services (driver writing, for example). I know
something about this as I work in a group which provides such services.
But that's what's neat about UNIX (oh I mean the UNIX OPERATING SYSTEM
(tm). Go ahead and use Ultrix as a noun for now... :-) ) -- it is an
operating system which can be held together by one or two so-called gurus.
And, as I would guess most of us have seen, a system which stays up for
long periods of times w/o problems. But, if you want to pay to have
someone else do these things (including updates, fixes, etc.) there are
people who handle such things.
Fred
--
Fred Avolio, DEC -- U{LTR,N}IX Support
301/731-4100 x4227
UUCP: {seismo,decvax}!grendel!avolio
ARPA: grendel!avo...@seismo.ARPA