I think Stephanes criticism of the free JVM landscape is quiet correct, and there is no clear way forward to resolve many off them...
However, I'd like to stress that there is a big distinction between a free JVM and free java software.
Java is a rich environment for free software (eg 1802 java projects on sourceforge vs 3616 C projects) and there are many "real programs" that have been written for JRE<1.3. (OK not many of them include usable GUIs but IMHO java is better suited to the server side).
Furthermore, while Sun can be criticized for many things, you have to acknowledge that they have developed a language and process that at least allow, more often positively encourage free implementations of java APIs.
So while the road to a free JVM may well be rocky and full of legal pot holes, it is a worthwhile one to go down. I hope that somebody does step forward to champion the cause in debian, if only to feedback to the free JVM developers where their projects are falling short.
As the developer of a major piece of free Java software, I can't take on this role myself but am happy to lend support, opinions, etc. to whomever does.
regards
Stephane Bortzmeyer wrote:
[If you want me to read your messages, copy them to me: I've unsubscribed.]
I'm changing, I'm leaving for a new employer and, in my new office, I will no longer use Java (which is a good thing for me, see hereunder).
Therefore, I stop maintaining java-common (I will send an official ITO unless someone steps in really fast) and the proposed Java policy.
One of the reasons I stop using Java is that it is too painful when you are commited to free software: most real programs depend on non-free (JDK >= 1.3, Swing). Even for those who do not, gcj and kaffe (unlike jikes) are, in their released versions, far from being 100 % ready. At least for kaffe, the problems of insufficient releases is worse in Debian because the package is too old.
Worse, many people in the free software world seem to care very little about the problem (the Apache group is a terrible example).
Another reason is the lack of standards in the way compilers and VMs are run, making the installation of every new jar a problem (defining environment variables, etc). The proposed Java policy tried to solve this and I would suggest that work on it resume. It is a tough job: everybody will disagree, few will suggest workable and proven solutions.
-- Greg Wilkins<[EMAIL PROTECTED]> GB Ph/Fax: +44(0)7092 063462 Mort Bay Consulting Australia and UK. http://www.mortbay.com
