[conspire] Christian draws out a Sun Microsystems guy

[Rick Moen]

Quoting Don Marti (dmarti at zgp.org):

> Red Hat made Eclipse run without Java.
> http://www.linuxjournal.com/article/7413

Red Hat guys have, I hear, been making exceptional progress in improving
GCJ and GNU Classpath.  They (along with other GCJ / Classpath coders)
are greatly to be commended.

There's something I'm a little unclear on:  My "Open Source Java on
Linux" notes suggested that GCJ's libgcj JRE is just an FSF-relabeled
copy of the GPLv2 "kaffe" JRE codebase.  However, more recently, some
comments I've seen online suggest that kaffe and libgcj are continuing
as separate projects.

So, my question is:  What's the story?  Did I get it wrong, and libgcj
was always independent of kaffe?  Was libgcj a _fork_ of kaffe?  Or was
it an independent project that borrowed from kaffe but didn't converge?

Either I was completely wrong about libgcj's origins, or there are some
details I'm missing -- or both.  ;->

> I predict a quietly "cleaned up" de-JREd release that runs a little
> faster than the official one.

The interesting question is which way the various distributions are
going to jump:  We could see a situation emerge where nobody but Sun
ships Sun's official proprietary-encumbered versions -- and everyone 
else ships a community-consensus fork that either snips the Java
portions entirely or includes only the portions that run reliably on
release versions of libgcj.

Further down the road, _maybe_ we could even see a community OO.o release 
that uses a beefed-up version of sqlite, instead of the Java hsqldb, as
the OO.o Base (integrated database) engine.

> RHAT and NOVL aren't both going to make their products
> dependent on the competition.

Good point.  I hope Novell's bright enough to figure out its
self-interest -- and maybe even bright enough to talk to other people
grappling with the same problem, for a change.  Historically, they've
seldom been good at the latter -- possibly a Utah thing.

One of the problems we've been having is that most people -- even in the
open source community -- simply haven't been understanding the problem. 
The Java guys produce a horrific tangle of jargon and TLAs that one must
master before being able to speak intelligently about the software:  It
took me a long time to get straight what JVMs, JREs, JDKs, JIT compilers, 
class libraries, Swing, AWT, SWT, and servlets are, and how they
interrelate -- and I'm still a little shaky on servlets.  Meanwhile,
people like Phipps are dazzling people with bullshit.

P.S.:  My LinuxToday comment on Christian's interview of Phipps:
http://linuxtoday.com/news_story.php3?ltsn=2005-04-28-013-26-IN-DT-SW-0022