cbsmith at envise.com
Mon Dec 28 17:52:57 PST 1998
Tom Gall wrote:
> Alexandre Oliva wrote:
> > On Dec 23, 1998, Christopher Smith <cbsmith at envise.com> wrote:
> > > I'll check which classes. I seem to recall it was the sun.io.* classes. I
> > > was using the version IBM provides to patch Microsoft's VM. Perhaps there
> > > is another version without such problems?
> > Maybe the same ByteConverter stuff that makes HotJava not 100% Pure
> > Java :-(
> It's more than that. Some of the implementation of RMI is native methods that
> needto do some inspection of objects as well as pickling and the like. These
> methods are not
> hard to write but you've got to be pretty knowledgeable in the full RMI
> as well as the deep down details of objects inside of the JVM.
> Sun has an RMI spec that has enough of the information in it one can figure it
> all out.
I'm pretty familiar with RMI, and I can't think of anything that
requires native methods. The reflective API's gives you solid knowledge
of any object, and the serialization API gives you 90% of what you need.
Furthermore, my understanding of the various RMI.ZIP's that are out
there for MS's VM are actually the same class files that sun was
More information about the kaffe