[omniORB] Complex numbers

[scott.rediger at teradyne.com]

>Will it really be noticable?  I would have thought that the cost of a
>simple memory copy would be dwarfed by the cost of sending these complex
>numbers through a socket, even for a tcp loopback or unix socket that
>doesn't generate real network traffic.  Perhaps you might notice a
>difference if the method invocations are within the same process, but
>even then the complex numbers are being copied a few times already (for
>marshalling / unmarshalling), right?

Agreed, the networking overhead is more than the translation memory copies.
However, he already has it working using pointer manipulation. If he
replaces that with translation functions, he may notice a difference in
performance.

ScottR