Hi!

wynand at realtimerodeo.net wrote:
> I don't know if this question has been posed before - I didn't have the
> time to search the archives for this, since I'm at a cyber cafe.

It probably has, but doesn't matter. :)

> When I first read about Squeak, I was mildly put off, since I thought it
> was an environment for kids. Now that I'm more farmiliar with it, it seems
> pretty robust - certainly enough so to serve as an environment for
> developing hardcore applications.

Yes, indeed.

> So, I was wondering, why is Squeak being marketed (inasmuch as one can use
> the verb "to market" for an open source project) almost only as an
> educational tool?

Well, it is not really marketed at all. :) Since the focus in the
beginning was for Squeak to serve as a platform for multimedia and
education, much of the "marketing material", like for example
www.squeak.org, was centered around that. After the original team (often
referred to as SqC - Squeak Central) dissolved nothing much has been
done in this department, so the material stands.

So the short answer is - Squeak isn't actively marketed at the moment.
But we could of course work to change this.

The future organization of the community and how we should work together
(and "marketing" is a part of that) is currently under quite a lot of
debate - I expect things to change in the near future - though I can't
predict in what ways. I can just say that I personally am working to
change it.

> Another question - is anyone still working on building a JIT component?
> I'm quite keen to do so, but I lack the experience and so I'll first have
> to do a lot of reading before I can start. Realistically it'll take me
> more than a year.

As Michael Latta wrote - contact Bryce Kampjes (often on IRC btw) about
Exupery. Exupery is damn cool and I played a little with it yesterday.
Unfortunately I didn't see the dramatic speed improvements regarding
bytecode execution that Bryce has on his machine, because I use a
Pentium-M laptop and... well, Bryce can explain the details. :) But it
was about 30% faster on bytecode. Bryce got about 400% IIRC, so 30%
wasn't much. :) But evidently there's a lot of stuff that can be done
and Bryce is AFAIK working on message send speed at the moment.

Generally I would say that Exupery is in my not-an-expert-on-JITs-eyes
the most promising thing in this area in a long time. Ian did a whole
range of Squeak JITs but even though his work was excellent nothing
generally useful for the rest of us came out of it AFAIK.

One aspect of this is that IIRC almost half of the execution of Squeak
(the larger benchmarks) is already being spent in primitives. And they
are of course running in "C speed". So even if you increase general
execution performance with say a factor 10 it would only amount to less
than a total factor 2 in practice.

Not sure if the above general "rules" still applies, nor if they are
valid in respect with Exupery, etc etc. In short - don't believe
anything I say. ;) And if we had Exupery we could perhaps recode stuff
without primitives/plugins blablabla.

Anyway, talk to Bryce.

> --Wynand

regards, Göran