Thu, 20 Jul 1995 21:48:44 -0400 (EDT)
In the misa discussions i think we're seeing closer on whats up with
generic functions but there are a few critical details missing in our
understanding that i will try to fix. i noticed that this seems relevant
to the tunes thread so i'm cross posting.
I agree with you completely that the foundation for a distributed object
system should be based on a message passing system of some sort. Generic
functions have the problem that they aren't "sent" anywhere so who does
the computation? who "owns" the gf?
however, i would argue that above a certain level of abstraction, gf can
be a very useful tool and should be in any complete object system. When
i said that gf have more "location" I was saying that from the
perspective of the caller. In my view, location is based on namespace
which is essentially a cluster of symbol objects. Whenever one refers to
a symbol, they are refering to a symbol object in their "workspace" or
now, the only sensible way to do generic functions in my mind would be to
have distributed definitions. there might be a "core" set of methods that
several people share, then to modify it, one would create a new gf which
would refer to the core gf with some new methods which would add to or
mask the core set, incrementally modifying its behavior. As i would be
refering to a generic function through my namespace, using my symbols, i
would refer to my copy which would then refer to the main copy. i present
a two level (core vs local) model here for simplicity but i'm sure it
could be easily seen how this could become a network model so that i
might refer to several different gf and add my own methods to modify it.
at the user level, this would be as easy as simply defining a new method.
It would automatically check for an imported reference to a gf then
create a new local gf, etc. You can't get to much more abstract than that.
refer to CLOS for specifics on how generic functions/methods are
At the McDonalds restaurant I did write on the restroom wall,
"For a good time call swell.hampshire.edu."