Re: Free Java, the sequel

[Mark Wielaard]

Hi,

On Thu, 2004-03-04 at 19:42, Tom Tromey wrote:
> Chris> Maybe we should write an open letter to the open letter writers?
> 
> If only we had some useful distribution channel.  Unfortunately I
> think it is too late to give a JavaOne talk.

It might be a good idea to express some of our concerns and hopes again
publicly to people. Although I doubt people who really care don't know
what the issues are with the way Sun and the JCP keep the main
proprietary implementation and "standards" under control and out of
reach of any free software developer. Luckily as you said we don't
really need Sun anymore. And I actually got asked on irc if we really
have to bother responding and doing "marketing" of our position instead
of producing more free code and making sure Classpath gets finished :)

That is btw what Bruce Perens told Dalibor and Chris when they explained
the open letters vs our free runtime environments.
He literally said: "Keep working on GNU Classpath."
http://lists.userlinux.com/pipermail/discuss/2004-February/004203.html

> It would be great to get our desires on the agenda though.  This is
> what Mark was getting at with the "SCSL and FSF" thread a week or so
> ago...

Sorry for not yet following up on that publicly yet. I said two weeks,
but I really want to get the 0.08 release out of the way first so I can
really concentrate on it (any day now!).

I actually started a discussion already about the way util.concurrent ->
java.util.concurrent (JSR166) worked and how/why cooperation with the
free systems didn't work out. This is really an interesting case since
Doug Lea who was the spec lead for that group really worked hard to do
it all in the open and in a way that the results would hopefully also be
usable for things outside the JCP/Sun JDK1.5 implementation.
http://altair.cs.oswego.edu/pipermail/concurrency-interest/2004-March/000888.html
(I also emailed with Doug in private about it already and in the past
both of us emailed Sun/FSF legal to try to work something out.)

Hope to turn some of the lessons learned from that into guidelines for
working with/without the JCP on standards. And analyze how/why the
JSR166 group and the results (public domain TCK!) was clearly so much
different (better) from much of the other JSR spec groups but still
didn't connect with the Free Software world.

We did drop the ball a bit in this case, since it is the closest I have
ever seen a JSR spec group work inside the JCP, but with free software
in mind. The licensing issues (interaction Public Domain/SCSL code
mixing) and the fact that some core results depended on one particular
proprietary VM implementation and the use of new language features, for
which no free implementations existed yet, did make it hard for us to
cooperate. Looking back we could have tried harder though. But it is
also a matter of trust, and I think some of us didn't really trust that
the results would be usable in free implementations in the end.

I am interested in other experiences of people who have (tried to) work
inside the JCP system on standards and/or implementations that could be
used with our free software systems. If you have stories about that
please email me or the list about it.

Cheers,

Mark

signature.asc
Description: This is a digitally signed message part