[omniORB] A newbie question (well, maybe more than one)

[Eric Damphousse]

Hi all,

I am very new to the wonderful world of CORBA, and have a nagging 
question that I cannot get a clear and definite answer for. I have in 
the past written an application that uses COM to the extreme, and have 
decided to rewrite all of it under Unix (Solaris to be more precise). 
That part is easy enough.

For my question. When I look at all the samples out there + reading the 
Advanced CORBA Programming with C++ (I just love C++) I see a common 
thread .. i.e. int main(int argc, char* argv[]). and always are those 
parameters used to call ORB_init(argc, argv). And my question is, is 
that really necessary. What I am trying to do is to have multiple 
objects that do a whole bunch of stuff, all accessed trough 1 client. 
All these objects would be built as shared libraries, not executables. 
Only the client would be an executable. Since all objects are libraries, 
one would not find a int main(blah blah). And hence my question. Is 
there another way of initializing the ORB if one does not have a int 
main(blah blah), or do all objects require to be executables.

This kinda begs another question, i.e., The client would have knowledge 
of the existence of all the other objects only during runtime. Again 
from the samples I have seen.. its always 1 client 1 server. I am trying 
to have 1 client and lots and lots of servers (objects). My guess, that 
is possible, but again have seen no proof positive that it is. The 
client I have in mind would only do one call to the ORB_init then use a 
IOR string to locate the server objects.

I am way off here, or is that all possible,

Cheers
Eric