Subject: 
             How to plug in new features in the RDF API
  Resent-Date: 
             Mon, 13 Dec 1999 16:42:34 -0500 (EST)
 Resent-From: 
             www-rdf-interest@w3.org
        Date: 
             Mon, 13 Dec 1999 22:51:55 +0100
        From: 
             Jonas Liljegren <jonas@paranormal.o.se>
          To: 
             RDF Intrest Group <www-rdf-interest@w3.org>




I am considering how to layer features as versions[1,2] and
probability[3] on top of the API.

 [1]
http://lists.w3.org/Archives/Public/www-rdf-interest/1999Dec/0067.html
 [2]
http://lists.w3.org/Archives/Public/www-rdf-interest/1999Dec/0081.html
 [3]
http://lists.w3.org/Archives/Public/www-rdf-interest/1999Dec/0085.html


The logic of this could be in another layer on top of the others. That
would be:

  Probability
     |
  Version
     |
  Schema
     |
  Simple
     |
  Source

Now I'm starting to think about if we maby should use some sort of
plugin, rather than more layers:


     Application
      /  |  \
     /   |   \
Version  |  Probability
     \   |   /
      \  |  /
       Schema
         |
       Simple
       / | \
      Sources 
     
But much functionality would change if you use higher layers. If you
ask for all atributes for a certain resource, the answer would be
diffrent if you are considering versions.


Should we overide the funcionality (by letting one layer inherit most
methods form the lover layer) or should we have new names for the
methods considering higher layers?

With the above layout, the application would inherit from both the
Version, Schema and Probability layers. Both the Version and
Probability layers inherits from Schema, since (or if) they are
dependent of the Schema layer.

But the Version layer should modify the functioning of the Probability
layer, and vice versa. Could they inherit from each other? How do we
solve this? (In perl, you can inherit from any number of classes.)


-- 
/ Jonas  -   http://paranormal.o.se/perl/proj/rdf/schema_editor/