Hi Geir, On Sat, 2009-01-17 at 06:40 -0500, Geir Magnusson Jr. wrote: > On Jan 15, 2009, at 7:11 PM, Mark Wielaard wrote: > > That is a very good point. Thanks for bringing that up. Currently we > > act > > as if the JCP has some kind of status that restricts certain kinds of > > modifications to public APIs. > > Yes, the JCP has *exactly* that status - public APIs must comply with > the Java SE specification as produced by the expert group, or else the > software is not Java (or, in deference to Sun's ownership of the Java > trademark, "Java compatible"). I think this reliable consistency is > one of the great things about Java The Ecosystem (as well as Java the > Platform).
Sure, consistency and having unambiguous, strong standards are great for everybody. No argument there. But if the rules around (producing) those standards are counter to the spirit of free software then it takes away from the user instead of adding value. We have to strive for public, open free specifications with testsuites that are free for everybody to use to verify any implementation's claim of compatibility with such standards. I am not convinced the current state of the JCP encourages that though. If through OpenJDK we can improve the process of producing specs, the reference implementation and free test suites, then I am all for it. > > But this has been kind of a problem since > > access to JSRs and JCKs is not guaranteed to be free of restrictions > > that are incompatible with our way of working in a public and open > > free software project. > > It turns out that's the least of your problems. Yes, the biggest problem was getting a full free reference platform for the Java platform. Although we worked very hard on that through the various efforts around GNU Classpath and friends, gcj, kaffe, and finally with harmony, it cannot be denied that Sun's liberation of almost all of their core platform implementation code base helped enormously. And doing it in a way that united their effort with almost all of the existing libre-java community can only be given the highest praise. I might be highly critical about some of the processes, the non-open specs, the TCK being non-free and only available under a NDA forcing people to work in secret cut of from the rest of the community (but again high praise for Sun coming up with something that at least lets people produce Free Software and doesn't get in the way of releasing the results under the GPL) and the non-transparent trademark rules. But I do realize that the biggest and most important hurdle has been taken now. That we are slowly but surely creating a community that produces a fully free Java platform together, even if some of the steps forward might be still tricky. > As you know, the ASF > is engaged in what is now a multi-year battle to get the Java SE 5 TCK > under terms compatible with being able to distribute the resulting > tested binary under an open source license. Yes, we started that process 12 years ago, and even before we started Harmony we tried to unite the free java groups and get access to old TCKs. http://lwn.net/Articles/184967/ One of the reasons that I was one of the Harmony founders, which I and lots of others hoped would be the ultimate unification of all the java-libre efforts that would not only bring us a solid, full, free java implementation shared by lots of groups, but also would give us the political cloud with the JCP community. We all know how that ended. As I said before I think your actions were not helpful. http://gnu.wildebeest.org/diary/2007/04/21/openjck/ I hope we can move past that sad history though and focus on the future. Now that we have a full free Java compatible platform for Java SE 6 lets focus on making the processes for getting the same for Java SE 7 and not just having free code, but also open processes (including open and free specifications and finally a free TCK!) instead of harping on the past failures. > Java will never really be free until we get past all of this. Please > inform RMS. You seem hung up on the term Java(TM). Yes, it would be great if we had a more open, transparent and Free Software compatible way of handling the trademark issue. But don't confuse naming with code. The code is all out there, under free software licenses. And even some binaries produced have been certified as passing the JCK - in a way that is less from ideal seeing the TCK itself isn't Free Software, but the resulting code is fully free software. I do talk with RMS from time to time and he knows my position, goals and the work that still has to be done. Cheers, Mark
