[omniORB] Complex numbers

[Marc Eberhard]

On Fri, Oct 10, 2003 at 07:32:37AM -0500, scott.rediger at teradyne.com wrote:
> Sure, that is another way to do it. However, your performance will be on
> the same level as xlate() because you are still doing object creation and
> translation.

Yes, sure.

On Fri, Oct 10, 2003 at 01:43:07PM +0100, Duncan Grisby wrote:
> And as long as the use of the template doesn't cause the C++ compiler
> to do anything unexpected.

Very true...

> Less dangerous than just casting types, since at least you're using
> the C++ type system rather than abusing it. However, what I had in

OK.

> mind was to dispense with the IDL generated Complex struct altogether
> and modify the call descriptors, object references and skeleton
> classes to use the STL complex type directly.

This would effectively break omniorb's compliance with the CORBA language
mapping to C++, wouldn't it? Is there some CORBA or omniorb way to create
such an extension to the language mapping without violating the CORBA
standard? Would you be interested in integrating such an extension into
omniorb?

> Regardless, like I said before, I don't think you should be thinking
> about any of this unless you find doing the easy translation function
> is too slow.

OK, I'll try that one first (probably in the form of inlined wrapper classes
instead of translation functions as I personally like that approach more).
I'll see, if it creates a significant runtime overhead or not in my case.

Many thanks to all who replied for the explanations and suggestions!
 
Thanks,
Marc
_______________________________________________________________________________

email: marc at greenie.net, marc.eberhard at alumni.tum.de, marc at affs.org.uk
email: m.a.eberhard at aston.ac.uk, web: http://www.aston.ac.uk/~eberhama/