[omniORB] Efficiency of omniORB 3.

b.keeping@ic.ac.uk b.keeping@ic.ac.uk
Wed, 17 Nov 1999 14:29:58 GMT


Duncan, on Monday you wrote:

> ... Assuming you're using omniORB 2.x, invocations to objects
> in the same address space are just C++ virtual function calls, so the
> ORB isn't involved at all. Local calls on omniORB 3 do a bit more
> work, but you still shouldn't see any problems.

The property of 2.x which you mention here is rather important to us:
to be specific, we are currently developing Corba components which
execute within the same address space for some intensive computations
-  probably hundreds of calls per second at least.

We haven't yet started trying to use omniORB 3, but could you give a
little more detail about the additional work per local method
invocation?

Ben Keeping
Imperial College