I'm trying to run trio on a linux machine that is also my house network firewall. The machine has a WAN 64.xxx.xxx.xxx address, and a local 192.168.0.0 NAT subnet.

When I try to load trio-applet.htm from the 192.168 address, it fails
quickly with a security exception in the browser plugin trying to connect back to the 64.xxx.xxx.xxx address. If I instead load the applet from the 64.xxx.xxx.xxx:8090/trio-applet.htm, it runs briefly,
then gets the exception trying to talk to the 192.168.xxx.xxx side.

FWIW, this thread at sun seems relevant:

http://forum.java.sun.com/thread.jsp?thread=311421&forum=58&message=1279662

but I'm still trying to comprehend it. The gist seems to be (a) it's complicated to make work right, with custom factories to get sockets that have -just- the right bindings; and/or (b) it might be best to avoid certain RMI callback patterns because they are prone to identifying source packets with the alternate host ip addre, messing up the security model.

It's making me wonder where I can find another machine to run trio on that is completely within the perimeter, but is always going to turned on. Right now, I'm only sure the firewall server is going to be up, and that's where I have my big disk of music. SIgh.

Ideas?

-dB

Actually, thinking about it, the real nut here is using an applet in the browser to begin with. This application might be cleaner if it ran all on the trio server side, sending pure html/javascript to the browser. It might not be quite as snappy, but it wouldn't require people to enable java in the browser which is also a bit of a drag. Applets seem like they were pretty much a dead end architecturally.

-dB

As tRio starts up, there should be some "binding RMI" messages near the top of the output. What do these read? If they're saying 64.xxx, then you can override this via the host-name in the <rmi> element in trio.xml:

<rmi create-registry="true" host-name="192.168.0.1"/>

The other way around this problem is to sign the applets. This will stop the security exception you're getting when you use the 192.168 address. If you'd like I can send you signed versions of the latest stuff - but be warning, it's not fully tested yet.

Cheers
Paul

it was binding to 127.0.0.1 and "localhost", not 64.xxx.
Trying to force it to 192, the result is the same -- it tries
to get to 64.xxx. I think the signed applet is worth a try.

I'd also be interested in some build instructions. I'm
a "C" guy, and the absence of a Makefile confuses me.

PS, I found making the [tune-name] a banner reasonably scolls
long song titles. So more than one banner at a time seems to work.

thanks,
-dB

I'm busy tracking down 2 problems that have been reported with the latest version and will then send you the files. Hopefully by the end of this week.

On the Makefile thing, the java code is built using ant (see http://ant.apache.org ). This is controlled by build.xml in the tRio directory.

There is a Makefile for the client-side C code (under riot in CVS), but you'll need a cross compiler to use this.

Cheers
Paul