[cap-talk] Objects and Facets

[Charles Landau]

At 5:34 PM -0700 8/7/06, Norman Hardy wrote:
>On Aug 7, 2006, at 4:39 PM, Mark Miller wrote:
>  > On 8/7/06, Charles Landau <clandau at macslab.com> wrote:
>  >> Unfortunately, we need some better terminology for describing
>>>  composites, which is why many people are using the term "object" when
>>>  they mean composite.
>>
>>  David's suggestion of "component" isn't bad. I haven't thought of
>>  anything better.
>>
>Component sounds like a part of something else, What is a component 
>part of?

Yes, that's why I don't like "component".

>I don't understand where you propose to use "facet" in the new scheme 
>of things.

An object/component is a collection of atomic objects. A facet of an 
object is one of its atomic objects.

>If you say that a facet is an object that holds a capability to the 
>underlying object, you are over-specifying the implementation.

We aren't saying that at all. In your counter example (not to be 
confused with a counterexample), we say that the Up facet and the 
Down facet share the counter, but we don't say how.

>My definition of the up-down counter said nothing about the nature of 
>the returned keys, except for the meaning of their invocation.

My definition was identical to yours except for one word (if we use 
composite or component) or zero words if we use my proposed 
definition of object.

>I can say an object bundles behavior(s) and state. What will you say?

A common and simple way to specify the set of atomic objects 
contained in a composite/component/object is to say it is all atomic 
objects that can access a given state. Such a 
composite/component/object bundles behavior(s) and state.