Introduction, new maintainer for kaffe

[Jim Pick]

> So, what does this mean for the Kaffe "custom edition", and the
> unification of it with the Desktop Edition announced on July 19th, 2000?

The "custom edition" was released as GPL sources with the PocketLinux stuff.
It is still on the PocketLinux website, somewhere.  I should copy the
tarball to the kaffe.org ftp site.  Tim was going to re-unify it, but he was
pre-occupied with other things (new baby, work), and seems to have lost
interest.  Nobody else took it upon themselves to do the merge, so it didn't
happen.

Since then, our internal version of Kaffe has become closed-source again
(although, fundamentally, it hasn't changed much since the PocketLinux
release).  It looks like we're going to re-launch it again as "KaffePro".
We'll have staff working on it, so it will probably diverge again from what
Kaffe.org has, particularily with regards to commercial support offered,
added subsystems, sub-licensing of patented bits, etc.

That said, I don't think Transvirtual has any interest in seeing Kaffe.org
be "crippled".  I'm going to make sure that if I find a bug in one version,
it's going to be fixed in both.  And I'm going to petition Transvirtual's
management for permission to release bits of code for use by the project if
those bits of code help kaffe.org along in any significant way.

> > 3) Clarify the relationship between Transvirtual and Kaffe.org.
> >    As a long-time kaffe-watcher, I would like to see Kaffe.org be a
> >    very open project, which incorporates code from, and interoperates
> >    with all the other free virtual machine projects out there.  I
> >    definitely see Kaffe.org as being an independent project that isn't
> >    controlled by Transvirtual.

> So, how does this fit together with moving all the resources under direct
> TransVirtual control? I'm just asking; I don't have a better solution. It
> seems that the perceived strong involvement of TransVirtual in Kaffe has
> led many open-source developers feel uneasy about it before, and with the
> recent pulling of the rug off the open-source Kaffe projects without
> warning or, in fact, even announcement still fresh in mind, I cannot
> imagine that attitude having changed for the better.

Actually, I like to think that's it under my direct control.  :-)

Let's face it - Kaffe was Tim's baby, and Tim ran Transvirtual and
Kaffe.org.  That's the past now.  What Tim did was create a Kaffe.org that
was separate from Transvirtual, even though he wasn't.  Now that Tim isn't
involved in either, Transvirtual isn't really in the position to try to
"control" what happens with all the code that's been released under the GPL.
The new non-Tim Transvirtual can either:

  a) be nasty, and try to destroy kaffe.org and the GPL'd kaffe (yeah,
right)
  b) keep kaffe.org alive, and ignore the GPL'd kaffe and hope it goes away
  c) actively support the kaffe.org project, and try to make it a success,
      hoping that people will see the support, and that they will take a
      chance at trying out Transvirtual's services and products

I'm not in charge of Transvirtual, but it's a small company, and I vote for
c).

I've been around free software/open source for a long time now, and I know
it's futile for anybody to try to "control" a project such as Kaffe.org.
These things have a life of their own.  If things aren't going well, and
there is interest, there will be a fork.  That's one of the things I love
about free software.

> >    As a commercial software company built on developing Intellectual
> >    Property, Transvirtual needs to be selective about what it does
> >    and does not contribute to the project.  You can expect that
> >    Transvirtual won't hold back bug fixes from the free version, and
> >    will not prevent others from contributing to the project.

> Assuming the KaffePro is based ont he custom edition, these two are
> already far too different for me to see significant synergy-advantage
> especially on the bug-fixes side, where the Desktop Edition due to its
> more open development model even has several bugs fixed that the Custom
> Edition didn't (And many more probably still exist). If improvements such
> as new JIT or GC modules are held back, soon enough there will remain no
> common grounds for the two projects to share bugfixes for.

I think the core VM's are currently quite different, that's true.  I want to
pick through the code of each, try to understand what's actually going on,
and formulate in my own mind what's good and what's bad.  I doubt that the
two versions have any differences that amount to a fundamental shift in
philosophy, as Tim had his hands in both of them.  Ultimately, most free
software and clean room Java stuff should be fairly similar, since it's all
just a re-implementation of Sun's stuff, following their fairly well laid
out specifications.

For example, our internal version a.k.a. the Custom Editition (PocketLinux
version) has a lot more support for JNI internally, I think.  That seems
like something that should be moved into the kaffe.org version.

The core class libraries (native,io,java.lang,java.security,etc) aren't all
that different at all, other than the native bits.  The AWT implementations
are completely different, but that's just a bolt-on part that can switched.
At work, I'm playing around with a new interactive configuration/build
system at work that makes it easy to switch components, provided they are
designed to be modular enough and the interfaces are clean.  Hopefully, I'll
be allowed to contribute that to kaffe.org.

Kaffe is more than just a VM - it's the also the core class libraries, and
the AWT (several versions) too.  I'd like to see swap the bits around, and
see how we work with competing implementations (ORP, Japhar, gcj, classpath,
etc.).  It might make sense to scrap entire subsystems and adopt third-party
bits.

What I'm most interested in is this:  The kaffe community has been around
for a really long time - anybody that's interested in this stuff know's
about it.  I think it has always had the potential to be more than it has
been.  I'd like to make it the place where people come to discuss and to
actually do the work to make free software Java work.  That means working
with other projects.  That means being open to contributions.

The actual design of the Java virtual machine is really cool, and using one
in the right place in a software architecture has a lot of advantages.  I
think it's a particularily good fit for people who want to write free
software, because of it's modularity and portability.  But up until now, for
a variety of reasons, Java hasn't been a popular choice.  I believe the use
of Java in free software should have been, and could still be, as popular a
choice as for doing free software as Python and Perl, if only it had a
similar sense of community.

> But then, there'll be amply of time to decide on issues like that once
> they crop up, just stating an observation for the record. I do understand
> TransVirtual's interest in maintaining their intellectual property
> private, and it has already given the "open-source community" what I
> consider the most complete and usable Java "clone" presently, for which it
> rightly deserves to be commended.
>
>  -Jukka Santala

Thanks for your feedback!

It gave me a good chance to ramble on a bit more on how I see things.

Cheers,

 - Jim