🚀 go-pugleaf

"Paul Murray" <pmurray@nospam.bigpond.com> wrote in message
news:qjaW4.1584$c5.2515@newsfeeds.bigpond.com...
> > Coming soon:
> > Run a Java GUI app under X. What strange parms will startx require?
>
> No probs.
>
> > Get an RMI HelloWorld happenning.
>
> Odd things are happenning. I set up a http server to serve up the classes,
> and set java.rmi.system.codebase to http://20.1.1.1:1234 on the server,
but
> the client on the other box doesn't seem to be even attempting to get the
> class form the codebase I specified. I know the server is working because
I
> can ping that adress and get some debugging output.

Yay! It's working - almost. My server sets up its own rmi registry and http
server to serve up the class files. I had some problems because I was
setting the codebase to
    ss.getInetAddress().toString()
rather than
    ss.getInetAddress().getHostAddress()

And initially the http server was attempting to return http header lines. I
probably got these wrong, but on the *client* it caused a JVM hotspot error.
So I'd say that RMIClassLoader on windows is
not doing the right thing when it gets bogus classfiles from the server.
Interesting.

Anyway, I removed the header lines and just did a dump of the .class file
alone a-la http/1.0

Now the actual object my server returns is a JLabel. When I run the client
on the local windows box there are no probs. but when I run the client on
the Linux box, the method returns a null. Hmm. I'd be prepared to believe
that a JComponent won't deserialize over there properly. I'll alter the
interface so that it's just passing strings around.

--------------------------------------------------------------
http://www.users.bigpond.com/pmurray
ICQ: 26066755