Re: javax.naming work (Classpath vs ClasspathX)

[Tom Tromey <tromey at redhat dot com>]

>>>>> "Mark" == Mark Wielaard <mark@klomp.org> writes:

Mark> I would not be against splitting up the standard/extension
Mark> libraries different from how Sun does it. But I don't think
Mark> splitting them between java.* and javax.* is better. If we
Mark> cannot come up with something better then I would suggest using
Mark> the J2SE as a way to define what we want to think of as
Mark> "standard/essential" libraries.

Partitioning libraries is an important question for libgcj.  I doubt
we want to compile everything available into a monolithic libgcj.so.

We do this currently, but as we gain mass we won't want that any
more.  For instance, I'm about to check in javax.naming.  Do we really
want that in libgcj.so?


This question is not completely identical to the question of where
libraries should be checked in though.  I'm personally in favor of
having as few upstream sources as possible, since that reduces the
things I have to do.  It also makes comparison scripts easier to
write.  And, finally, it makes it simpler for gcj contributors.

That said, I don't think it is a very big problem if there are
multiple upstream sources.  We already have 3 (zlib, Classpath, and
the GC).  One more isn't going to be a killer.

Tom