Re: Cobra: ILU or Orbit?




"Greg S. Hayes" <sdc@choice.net> writes:

> > ORBit works.  We have total control over its implementation (unlike MICO and
> > ILU).  It's lean (MICO wasn't).  No compelling reason to change.  Not to
> > mention that if we did switch to ILU, we'd have to throw out all the intensive
> > work that's gone into ORBit.  Makes no sense to re-evaluate again at this
> > point.
> > 
> 
> 	Will we be able to gleen some of the programing language interfaces
> from ILU (C++, Java, Guile Scheme, Perl, Lisp, Python, etc) or is the
> ORBit code so drastically different that this would be impossible?

It's completely impossible. ILU has it's own system of multiple
language support that doesn't really map onto CORBA very
neatly, and wouldn't match the POA at all. Also, the internals,
are quite complicated, and use a lot of ILU-specific abstractions.
That doesn't mean that some inspiration can't be gained from
the way I do things. The ILU Python mapping was my main
source for my ILU Perl mapping, and that in turn was very
useful in doing my Perl module for MICO. 

In fact, that was my general impression of ILU - it's a pretty good OO
IPC system, and very well implemented in some some ways (the biggest
advantage it has over ORBit is that it is designed for multi-threaded
operation from the ground up), but it's only approximately CORBA.
Trying to pretend that it was exactly CORBA was a somewhat
frustrating task at times.

Also, out of the ILU languages, the Perl binding is pretty much
abandoned (I can say that authoratively ;-) and I believe that the
Guile support is in a similar state. (Actually, the Guile code was,
last time I looked, based on a custom-rigged object model, and looked
in need of major revision to something more standard.)

Regards,
                                        Owen




[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]