Is dolphin dying ?

mario bancos wrote:
[snip]

> On the other hand, don't you think the (cheaper) VW per year licences is
> a marketing decision to win in the low budget segment of the Smalltalk
> market, where Dolphin, MT and VS (another St. ditribution killed by
> Cincom) are the principal rivals.

This is almost funny, but also deeply mistaken.

<rant>
First VS was killed by ObjectShare, not by Cincom; Cincom has done their
best to make VS available, including maintennance releases, e.g. for
WinXP; i.e. the VW team puts effort into keeping VS alive, even if in a
coma.  Further, note that some of those behind the "killing" of VS at
ObjectShare were senior figures at Digitalk.  VisualWorks did not kill
VS, people in control of PPD/ObjectShare who were interested in making
money killed it, and they ended up loosing the company a lot of money,
and a lot of engineering talent in the process.

One major mistake ParcPlace and Digitalk management made was assuming
they were each other's competition and hence a merger eliminated each
other's major competitor.  Java was just a few months away...

Second, Dolphin is not VW's competition, it is our ally.  .Net and Java
are our competition (and to a lesser extent python, perl et al).  It is
to compete against MS, IBM, Sun et al that we've provided a lower
entry-point for the single programmer, not to steal market share away
from Dolphin.

In fact the situation is quite the reverse.  VW wants Dolphin to succeed
and we're worried whenever any dialect seems to be suffering.  If one
looks at Smalltalk from an MIS perspective it can be perceived to be a
tiny niche with very few stable players.  Market analysts denigrate
Smalltalk as a dead language that is going away and one that management
should leave in favour of typically Java and .Net.  [But these analysts
are playing to the choir, not providing objective advice].

The more healthy vendors and active open source dialects there are in
the Smalltalk community the more the above misperception can be
countered and the more confident MIS types can be in choosing Smalltalk.


If one looks at market share as available dollars to be spent on
development and deployment technology then the choice is obvious.  One
can wear blinkers and go after 100% of the few millions of dollars being
spent on SMalltalk development projects, attempt to eliminate the very
people that help bolster your own sales, and have a larger slice of a
rapidly shrinking pie.  Alternatively, one can look at the total market
and attempt to gain a share of billions of dollars being spent on
development and deployment technology across the industry, and gain a
growing share of a growing pie.  For the Smalltalk sector to to do the
latter it helps if it attempts to be a community and recognizes its
members can be of enormous help to each other.  Being fearful of each
other is not the answer.

One important aspect of this is the evolution of Smalltalk.  When
Smalltalk was developed it was funded by the most rapidly growing
technology company the world had seen and it was developed by a
relatively small team.   Certainly the number of people working on Java
at IBM and Sun dwarf the amount of people working on hardware and
software at PARC in the 70's.  i.e. it was relatively cheap for Xerox to
fund in the 70's, but funding a successor now would be much more costly
if Java and .Net are at all representative (which they may not be).

Now, if Smalltalk is to evolve, or a successor invented to obsolete it,
I think it extremely unlikely that this will happen in the context of a
corporate funder.  i.e. I doubt that Alan Kay will be able to get HP to
provide sufficient commitment to do this.

Where else might it happen?  The two obvious candidates are in
universities and in the "open source community".  But since it is
universities that populate the open source community anyway we should
concentrate on universities.  That is a place where people get exposed
to new ideas, fall in love with them, and often come up with good new
ideas.  Companies like MS recognize this, which is why they are
targeting universities with technologies like Rotor (the open source
.Net platform) and funding for research.  They are fighting for hearts
and minds.

Over the past two decades the university sector has become more
vocational in its teaching.  Alan Kay lambastes no less than Stanford
university in his Croquet presentation for using Java for teaching.
When I was teaching in London University in the early 90's much debate
was between those that wanted to teach concepts and those that wanted to
"provide marketable skills".  Government, with pressure from industry
(almost always short-sighted), sided with the vocationalists and good
computer science teaching suffered.

So if universities are to be places where people get exposed to the good
stuff like Smalltalk, Lisp and Prolog, one thing that will definitely
help is if the commercial members of these communities can demonstrate
that in fact their technology is not dead, is not esoteric, but in fact
in widespread and extremely demanding use in industry.  [side note:
VisualWorks and VW/GemStone combinations are used in sectors such as cpu
manufacture, container shipping and derivatives trading on a world scale
(i.e. they handle a substantial fraction of the world's activities in
these sectors).  But for nearly two decades the corporations who have
built these applications have viewed their use of Smalltalk as a
strategic advantage, and hence prevented the vendors from using the
applications in marketing material.]

The more the Smalltalk community can demonstrate commercial viability
and relevance the more widely  it will be adopted by the universities
and the more minds will follow the arc of falling in love with
smalltalk, finding its limitations and dreaming of something better.
</rant>
--
_______________,,,^..^,,,____________________________
Eliot Miranda              Smalltalk - Scene not herd

kuo wrote:

>    Totally agreed. Seaside is a wonderful innovation but
unfortunately stuck
> deeply in the dialect-specific environment, IMHO, it depended too
much on
> Squeak's unique VM mechanism, AFAIK, it was almostly tangled inside
it, you
> had to be quite familiar with Squeak so that you could extract it out
from
> Squeak faithfully  to port successfully to other dialects without any

> distortions.

This was arguably true at some point, but these days we're actually
quite careful to keep Seaside dialect indepedendent.  I don't know
exactly what process Michel goes through on every release to do the VW
port, but from what I understand it's almost entirely automated;
certainly, he tracks the Squeak version extremely closely.  Every once
in a while we catch something that gets in the way of portability
(usually stuff like assuming #foo = 'foo' is true), fix it, and move
on, but it doesn't happen that often.

I'm looking forward to an active Dolphin port to keep us even more
honest in this area.

Avi