🚀 go-pugleaf

HADAR PEDHAZUR
--------------

The main speaker at the August 12 GUN meeting was Hadar Pedhazur, from
the Union Bank of Switzerland.  He spoke of 3.0, related the history
of his involvement with NeXT, and gave some advice on developing apps
in the financial marketplace.

3.0 IMPRESSIONS

Hadar thinks 3.0 is very solid, from the user's perspective, and
expects to upgrade his user machines to 3.0 as soon as he gets it.

In 3.0PR1, he was disappointed with DBKit, but with 3.0PR2, he has
more hope.  DBKit does much of what businesses want: simple sorts,
searches, joins.  People who do very sophisticated things with
databases may run up against DBKit's limitations.  Nevertheless, Hadar
said DBKit beginners were doing impressive things every day, and
developers are busy writing adapters.

He looks forward to using distributed objects a lot.  At the moment,
the NeXT computer is not the very fastest machine in the world, but
with distributed objects, that may not matter, because the work can be
spread around in new, creative ways.  He's tried distributed objects.
They're real, they work.

As for the IndexingKit, Hadar thinks it's a great idea, but he's
annoyed at the changes in Digital Librarian (between 2.x and 3.0PR2).
DL's summary lines sometimes contain no useful information.  [Also, DL
still crashes from time to time when multiple targets are selected.]

HISTORY

When Hadar was introduced to the NeXT cube, he thought it was too
slow, and rejected the platform.  Because he thought NeXTstep was
somehow more special than anything he'd seen before, he went to Dev
Camp anyway--years ago--and was lucky to see a prototype 68040 that
was 3x faster than the cube, so he stuck with NeXT.  At Dev Camp,
Steve Jobs pulled a prototype slab out of a black bag.  Later it
turned out the slab was made of wood, temporarily.

Hadar makes Program Trading and Equity applications.  According to
Hadar, programming financial services apps is "not rocket science,"
but sometimes you have to be clever, because timing is so critical.

The biggest advantage he sees to NeXT is rapid development time.  He
also has numerous programmers who use NeXT computers successfully
without knowing Unix--they concentrate their thoughts on the appkit.
He says this is not be possible on a Sun (you have to know Unix).

The longest he's seen it take someone get up to speed on NeXTstep is
two months.  That person is now a super programmer.  He's also seen
people get their arms around NeXTstep in two weeks.

ADVICE

Hadar recommends FAME for any kind of time-series database work.

When writing object-oriented software, don't try to design the perfect
object hierarchy on your first pass, or you'll never finish your
application.  You really learn what the right objects should be
through trial and error.  Be patient.  Get something working first.

Let your programmers play with the machine and get a feel for it.
Initially, they will write ridiculous programs that bear no
resemblance to what you want.  Don't worry.  Soon enough they will
turn play into productivity.  Again, be patient.

He likes the Perl programming language, and uses it a lot.

A member of his group wrote an application for bank XXX in three
months using NeXTstep, and this application is still in use.
Meanwhile, XXX dumped NeXT, and nearly wasted six persons times 1.5
years developing a Smalltalk-based application on a Sun, which turned
out to be a flop.  [This is interesting, because another institution
is considering a similar approach.]

On the topic of Smalltalk, it is nifty, but it also wants to own your
machine.  If you make any calls external to the Smalltalk environment,
expect your machine's performance to drop precipitously.

Hadar related an interesting training story.  Two programmers from
London came to New York, to verify that the wonderful NeXT custom
applications they were seeing were not so much fluff.  They spend two
days, Thursday and Friday, just watching New York programmers go about
their work, i.e., using Interface Builder, Digital Librarian, Emacs,
gdb, etc.  The Londoners brought their PC apps on floppy, and on
Saturday and Sunday, reproduced their PC applications under NeXTstep.
Hadar spent two hours getting them "unstuck" now and then, but that
was it.  The Londoners were delighted with the four-day experiment,
and Hadar learned an important lesson: that the promised NeXT
productivity is there if you can somehow make sure a resource is
available to unstick a momentarily stuck mind.  With one mentor
around, newcomers can become productive quickly.

GEORGE CUMMINGS
---------------

George Cummings, in charge of Sales for the Northeast United States,
spoke next.

J.P. MORGAN

He started by commenting on the recent news that J.P. Morgan had
decided not to standardize on the NeXT platform at this time,
preferring a combination of Sun and Macintosh computers.  George said
NeXT will win back the Morgan account, but it might take a
while--perhaps a year.  He said part of the problem was that there
were 25 NeXT programmers at JPM, against 300 Macintosh and 250 Sun
programmers, and that JPM had a lot of existing investment in
architecture to consider.  [JPM also commented in a securities
publication that it was unsure of the portability of Objective-C.]  On
the good side, he said that JPM fully appreciated the technical
capabilities of NeXT computers.  The loss of JPM is a blow to NeXT,
but George said NeXT will reveal several buyers of equal stature to
JPM in the next 90-120 days, so keep the faith.

3.0 -- 3.0 will ship in two weeks (end of August).

4.0 vs. more NeXTstep ports

Numerous persons are planning 4.0 at NeXT, where 4.0 has a higher
priority than porting NeXTstep to more hardware platforms.  The reason
for this is economic reality.  Support for lots of platforms is
expensive, and NeXT would prefer to continue innovating.  If money
were no object, though, NeXT would spend more resources porting
NeXTstep to new platforms.

NeXTstep-486 -- NeXTstep-486 is expected to ship in 1Q93.

NRW -- NRW apparently is a code name for new hardware from NeXT.
George kept mum about this, except to utter the party line, which is,
"NeXT will never fall behind the industry's performance curve."

DATA GENERAL

NeXT has recently partnered with Data General.  The purpose of this is
to make use of Data General's extensive sales force, 1200 strong,
which reaches many markets new to NeXT.  DG would like to promote the
idea of NeXT computers on the desktop attached to DG's Avion servers.
The marketplace likes Avion servers, but has not embraced Avions on
the desktop.

SUN

George said that Sun isn't all bad.  For one thing, Sun is doing a lot
to legitimize Unix on the desktop.  [See NT below!]

As for porting NeXTstep to the Sun SPARCstation, apparently Steve Jobs
said NeXT would do the port "if `they' will pay for it."

MICROSOFT NT

This wasn't said at the meeting, but after dinner later in the
evening.  Two experts I respect have said very positive things about
NT.  The scariest thought, according to one, was that NT could be a
"Unix killer."  In other words, Microsoft may have finally achieved
technical sophistication.  Steve Jobs has called NT a "Unix wannabe."
But if he's wrong, imagine a world without Unix, and try to figure
your place in it, or what you'll do about it.

-----
(c) 1992 G.S. Knauth <gsk@marble.com>
standard disclaimers, copy permission granted for user groups
-- 
Geoffrey S. Knauth, Marble Associates, Inc.                Member BCS-NeXT, LPF
gsk@marble.com, (617) 891-5555 work, (617) 547-5247 home   Standard Disclaimers