(unknown)

[...]
> Translation+execution=interpretation => we are speaking about the same
> thing.

Ok, I'm with you :-)

[...]
> > > I would not worry about different users. Each one of them would
> > > simply use a subset of all possible requests.
> >
> > But someone has to check whether they are authorised to use the subset
> > they are using. This is overkill in the simplest kind of applications,
> > but it is definitely needed for commercial components.
> >
> 
> OK. This is new - here we are speaking abot authorisation. This can be done
> by upending to each request the code of the category of user. Using that
> code the component can see if the request is a valid one from that category
> of a user.

Something like that, yes. But authentication is (will be) something
that can be outsourced to the CORBA implementations, so I think it's
the interpreter's task to know where authorisation can be done; he
need not do it himself. So, again, not too much code bloat to be
expected here.

Other things that can be outsourced to (future) CORBA implementations
are the persistent saving of the component's state (when necessary),
management of transactions (do we need those?), and a couple more.
See <http://www.orocos.org/corba.html> for a list of these `CORBA services'.

Herman

-- 
K.U.Leuven, Mechanical Engineering, Robotics Research Group
<http://www.mech.kuleuven.ac.be/~bruyninc>  +32 16 322480