SysVR4?
Carl S. Gutekunst
csg at pyramid.pyramid.com
Wed Sep 21 04:01:57 AEST 1988
[Mandatory disclaimer: What follows are my very own personal opionions. I have
neither the authority nor the knowledge to represent Pyramid Technology Corpo-
ration on any of the following issues. My comments are *not* based on inside
knowledge of anything going on at Pyramid; just my own observations based on
12 years of experience in the computer industry, and five years with UNIX.]
In article <856 at galaxy> dave at andromeda (Dave Bloom) writes:
>Though I've been a great fan of Pyramid's OSx, I'm glad to see the whole
>unification effort going on at AT&T/Sun.... it looks like the usefulness
>of the dual universe concept is finally coming to an end.
That would be nice if it were true; OSx is a pain in the wazoo. But it's not
going away, for a lot of reasons.
- Sun != BSD. Good things continue to happen at Berkeley -- better than what
is coming out of Sun of late. As long as Berkeley continues to develop good
things, customers will ask for them, and we (the developers) will want them
too.
- SVR4 is like one of those patchwork monsters in the old Japanese movies. A
piece here, a piece there, all glued together. It's big, crufty, unmaintain-
able, and slow. File System Switch *and* V-nodes? Wollongong TCP/IP? Gag me
with an Ethernet transceiver! Note too that AT&T is only taking the very
best of SunOS, those things that make the biggest splash. Lots of smaller
things they are doing themselves, and *not* doing them in ways that are
compatible. Guy Harris's streams TTY driver is a beautiful bit of work, but
AT&T rolled their own, and I hate to think what it looks like.
- Merging is still a long way off. SVR4 and SunOS 4.0 are evolutionary, just
another step in a long process. Note that SunOS has lots of dualport-like
hooks, ways to give the user the choice between System V semantics (so they
can come closer to passing SVID) and BSD semantics (which is what all their
customers are used to). As long as Sun has to have a dualport mechanism, so
certainly will Pyramid.
- SVR4 still smells like System V. I don't like System V. Neither does the
vast overwhelming majority of Pyramid's customer base. They buy because of
the att universe. They live in the bsd universe.
- I am increasingly coming to agree with the "Hamilton Group" assertions. AT&T
and Sun will exchange the first and the best of everything with each other,
and the rest of us will get the crumbs.
- A lot of us have been predicting Sun would reach critical mass and explode
"soon." Meaning: crash and burn, in the glorious style of Apple, Tandon, and
AMD before them. All the people who have stayed with Sun because of their
"golden handcuffs" take off, and the wonderful pool of talent that Sun built
up (and that AT&T is currently bleeding dry, just like they did to the good
people at Convergent) will take off for greener pastures. This was averted
this year because of the infusion of cash from AT&T. But that's just delay-
ing the enevitable. Either that, or Sun becomes a subsidiary of AT&T, which
would make all the good people leave, too.
As I've groused before, I think the whole Sun/AT&T deal is a hideous waste of
time and programmer resources, and OFS doubly so. I just don't see any evi-
dence that either new or future customers are demanding a merged UNIX, while
there are lots of other things that *must* happen for UNIX to be a commercial
success. It's not yet, you know. We in our little corner of the computing
world are terribly naive when we assume UNIX has the world by the tail.
(It was fun discussing this with people at the PUG conference. A lot of you
agreed with me, and a lot did not. Nothing happened that changed my opinion.
The claim that I am blinded because Pyramid "has a solution" in OSx is not
true; OSx is *not* a "solution" to the UNIX incompatability issue. It is a
stopgap measure, and a very effective one. But it is not a solution.)
>The representatives from AT&T/Sun intimated that there will in all likelyhood
>be a number of vendors SHIPPING Rel 4 within WEEKS of AT&T's Release in 3Q89.
>Will OSxV.4 be one of them????
That one is easy, even if I wasn't working here. Of course not. Pyramid has a
tremendous investment in its installed base, and in all the support products,
like communications. Switching operating sytems cannot be done trivially. Odds
and ends -- useful ones -- from SVR4 will gradually be folded into OSx, but we
sure aren't going to abandon the entire product line next year. There are lots
of much more important things to do, and programmer resources are finite.
I honestly have no idea what Pyramid is planning with regard to SVR4. Sure,
the company is looking in to it. But for myself, if Pyramid were to persue it
seriously as a replacement for OSx, I'd make every effort to persuade people
that this was a bad idea.
As far as vendors shipping SVR4, it will likely be similar to the majority of
vendors who've signed up for SPARC: startups, or companies with a history of
failure in the computer industry who are looking for an easy way to buy back
into the game. Oh, and probably chip manufacturers, too; I would expect Intel
and Motorola to have SVR4 ports as soon as they could.
<csg>
More information about the Comp.sys.pyramid
mailing list