[Zope3-dev] Yet Another Relations (aka Reference) Engine...

[Jean-Marc Orliaguet]

Helmut Merz wrote:

>
>OK - then I'm not talking about references, though we still mean 
>different things when talking about 'relation':
>
>Mine goes like this: A relation object R providing 
>ITriadicRelation and connecting A, B and C involves three 
>references: 
>
>R with A
>R with B
>R with C
>
>In addition the relation object has a sort of type, the 
>'relationship' (corresponding to a 'predicate') that may be 
>specified by the relation's class or - a special case - a 
>string.
>
>Maybe there is a better word for this kind of 'relation'?
>
>Helmut
>  
>

I don't know. I think you have to start from a definition that is not
dependent on zope in general and understand what the concepts mean
outside the computer world. This was not invented here.

The "relations" used in software are more than just relations since they
also assert something, and a relation by itself asserts nothing, it is
just an abstraction that connect several things or aspects of things. So
"relations" here are rather "propositions".

for instance:
"A blue box" involves a relation between the "box" and the quality
"blue", but to represent the relation, you would to assert a proposition:

    "The box is blue"

The proposition can be analysed. It has a predicate and a subject (or
subjects if taken individually), the subject is an ordered tuple and the
predicate is what is left of the proposition when the subjects have been
removed (it is what is "predicated", or asserted of the subjects of the
proposition)

for instance in the proposition:

    "The box is blue"

the predicate "__ is blue" predicates "blueness" of the box

but a predicate is not the same thing as a relation

/JM