Michael Robinson wrote: > > >I was wondering if this was the intended overall approach? > > It was my personal intended overall approach. I want to use Python as a > middleware/second-tier scripting language. In those applications, you > almost always have a well-defined interface (or, at least, should), and > performance is very big consideration. > Ah, right. My motivation is simply writing objects for desktop use, and maybe running python objects in the same address space as C++ objects and perl objects if there are shortcircuit performance benefits to be had. This would mean that the dynamic interface to the orbit core would be an abstraction above IIOP - maybe GIOP compliant, which the orbit core could then marshall onto the IIOP wire, or directly to other objects running in the same orb. > I can appreciate that for other applications people would want dynamic > marshalling. I think the world is big enough for both approaches, and > people should work on implementing whichever approach they have the most > pressing need for. Agreed. Cheers, Phil.