Term "Configuration"

Brian T Rice water@tunes.org
Sat Apr 19 19:22:01 2003

On Sun, 20 Apr 2003, Jeff Cutsinger wrote:

> I've been lurking in this discussion for a while, and i want to make
> sure i have the idea correct. Basically, a configuration is a collection
> of objects + a set of rules about how they relate or must relate to each
> other?

Well, we have to have rules that are actual objects. The point of
configurations is to refer directly to the things at-hand, so I can't use
equations unless they are expressed directly as meta-objects of the given
ones. But otherwise, you are correct.

> Brian T Rice <water@tunes.org> said:
> > To return to this interesting topic (to me), what do people think about
> > characterizing multiple-argument functions as Attributes of
> > Configurations? This was one of the senses in which this concept occurred
> > to me.
> >
> > Bring up examples from your favorite odd language that don't seem to fit
> > it. We could judge by applicability.
> is this similar to the way TOM handled them? TOM used tuples to return
> multiple values, so a function might return say (int, bool, int), and to use
> these values one would write:
[Omitted return-multiple-values example.]

Yes, that's similar, although the collection, you see, is implicit, and
the primary return value is what is received if the caller does not
explicitly grab the multiple values. At least, this is what happens in
Common Lisp. I am unsure about TOM's case for handing this.

What we could do in a more general sense is to not have to rely on
arbitrary structuring. Basically, we could have a convention to take in
and return configurations. There might be a further convention for a
default member of the configuration. The other parts might be attributes
of the first value, or they could be both attributes of something else
(which is somewhat similar to an array or list). If the configuration is
more abstract, there would need to be a meta-object relating them,
although I suppose this could be an attribute via isMemberOf or something
like that.

There's also the case that the values are attributes of the computation,
such as 'dummy locals' for "quotient" and "remainder" for integer

Your multiple bindings example could be done via expressing a
configuration of quoted terms, perhaps. I'm not sure what the correct
analog is; perhaps the configuration itself would be quoted. Or you could
simplify this by serializing the configuration and using a list of
variables (attributes) to bind to them.

Brian T. Rice
LOGOS Research and Development