[omniORB] using multiple ports

[Renzo Tomaselli]

Kendall,

> Faced with the same problem, I think I would write a special port finding
> application, such that it reads the application name on stdin and writes
> the official port on stdout.  The real application can then spawn this
> (popen) prior to initializing it's orb and not go through two orb
> creations.  Using omniORBpy, the port finding app might take only a few
> minutes to write, and would be easy to maintain.

I didn't mean to create multiple ORBs. In fact, this is not possible nor
needed. OmniORB can be inited multiple times, where the very first one it's
a real initialization (e.g. the orb is created). All the others retrieve it.
What I'm not sure about, is the possibility to add a new port to listen at
*after* the initial orb creation. And AFAIK, init/shutdown/init is a
forbidden sequence.
Beside that, we have a pretty complex architecture, made up of distributed
dll/shlibs, dynamic loading on-demand and such. All of this under an
(OmniORB based) security system. Nothing of this complies with a standalone
application to retrieve a port and tell it around by pipes.
Thanks,

Renzo