System V Release 4.0 Developer Conferences

[Henry Spencer]

In article <8570 at smoke.ARPA> gwyn at brl.arpa (Doug Gwyn (VLD/VMB) <gwyn>) writes:
>>A friend who attended said it was horrifying, actually...
>
>Come on, Henry, how about some INFORMATION.

Herewith roughly what my friend told me.  Note, this is him talking (more
or less, I did the typing based on my notes from a phone call to him -- he's
hard to reach by mail), not me.  I take no responsibility for errors! :-)

---------
The biggest problem is "filling in the gap rather than narrowing it".
"Why do you want both NFS and RFS?"  Many people saw this as pointless
duplication of effort, leading to unnecessary complexity and all the
problems that brings.

Details on licensing were scarce.  Worse, direct questions on the subject
were evaded rather than answered.

There is no agreement between AT&T and Sun to avoid divergence of the
two systems *after* SVR4.  This horrified everyone.  "Then what's the
point of all this?!?"

The kernels will be very big, probably needing a 16MB machine just to get
started.  Nobody liked the sound of that.

Administration will be a nightmare, especially with two different network
filesystems and all the permission complications that implies.  What are
the semantics and implications of symbolic links in such an environment?

Training people to use and administer such a complex system will be
lengthy and difficult.

There was inadequate detail on the commands that will be part of the
standard environment.

ABI implies that a conforming machine must have kernel support for both 
NeWS *and* X.  This is ridiculous for people who plan to run neither.

The over-friendly system-administration interface isolates even the
knowledgeable administrator from what is really happening underneath.
This is fine IF IT WORKS ONE HUNDRED PERCENT.  People with experience
with things like 3Bs know damn well that it doesn't, at least not as
normally supplied by AT&T.  And when it doesn't work, you're up the creek,
since the internals are not documented.  Troubleshooting one's own
problems is impossible.

In general, overall, the system is not getting simpler or cleaner.  Quite
the contrary:  it is getting bloated, complex, and slow.  It will be
difficult and expensive to run, support, and learn.

As for the presentation, it was good but not great.  The level of detail
was inconsistent and often ridiculously low for software developers.
"Two hours one morning when we learned nothing whatsoever."  "Much more
technical content could have been provided in the same length of time."
People did *not* like the way some questions, e.g. licensing, were evaded
rather than answered; even an honest "I don't know" (which is what the
real software people normally said in that situation) is better.
---------
-- 
NASA is into artificial        |     Henry Spencer at U of Toronto Zoology
stupidity.  - Jerry Pournelle  | uunet!attcan!utzoo!henry henry at zoo.toronto.edu