Exposing COM via XML-RPC or Something Else

[Gordon McMillan]

Bernhard Reiter wrote:

[snip a huge pile of links]

> >CORBA is NOT a component standard -- it's an interoperability
> >standard.
> Well AFAI understand it COM also is an interoperability standard,
> but I might just miss the point here.

"COM" is many (too many) things. At it's core, though, COM 
is a C++ vtable exposed to C. In that sense it is a binary 
standard, and very simple.

COM development followed two conflicting tracks. The C/C++ 
folks followed one track, and the VB folks took another. The 
VB stuff is complex (because the VB folks took all kinds of 
shortcuts based on VB internals). It's also what caught on 
(IDispatch, Automation, the stuff the Python's COM 
extensions do so well). The equivalent CORBA stuff (dynamic 
discovery) is probably better architected, but nowhere near as 
widely used. The part of CORBA that is widely used is 
actually much more straightforward in COM.

> >>The big part of the microsoft COM platform is the MTS
> >>(microsoft transaction server) if you want to do to distributed
> >>objects.
> >
> >Um -- that's true for COM+, but COM doesn't need it to operate.
> Well if you don't want distributed computing...

MTS provides transaction services. That's not part of core 
CORBA either. Distributed computing is DCOM, which is 
COM plus some marshalling.
 
> I do not habe much experience about it, I admit, but technically
> they also seem to have the same complexity. I just do not believe
> that COM is easier to code. Well this is, what the articles
> support, too.

It all depends on what you're doing and what tools you use.

CORBA was a full blown spec in 89, but didn't have any 
implementations until 95 or so. COM was working in 90, but if 
it ever had a full blown spec, I must've missed it.

- Gordon