Re: Corba-izing gnome-pilot

[Manish Vachharajani <mvachhar vger rutgers edu>]

On Thu, 19 Nov 1998 jrb@redhat.com wrote:

> Hi,
> 
> I am working more on the address conduit -- mostly trying to get
> gnomecard CORBAfied.  I would like to change the API a little.  When
> the CORBAfied branch of gpilotd gets finished, we should pass in a
> CORBA_ORB to conduit_get_gpilot_conduit.  Each conduit should not
> create it's own orb, but use the pilot's.  In addition, gpilotd should
> register itself with the name server so the capplet can find it when c
> 

I am not too familiar with how ORBit works, but if we don't pass this
parameter, will the conduit just create its own data structures to
communicate with the orb or will it actually instatiate another instance
of the ORB?  I don't want to make conduit development rely on ORBit
devel stuff being around if we can help it.

If it creates a new ORB process we need to find a way to prevent this.  If
there is no other way then I guess we have to modify the API.  If the a
new ORB process is created, does that mean that each gnome application
that uses CORBA objects in separate memory space uses a new ORB???  How do
the two ORBS communicate then, through IIOP?  It seems that this is a
little inefficient, are there any reasons this can't be avoided in the
ORB itself.  Hmm, maybe I should look at the ORBit code :)

Manish

Manish Vachharajani               Some Haiku: A crash reduces
<mvachhar@vger.rutgers.edu>                   your expensive computer
                                              to a simple stone - Unknown