It took a few days to think about this new model. I think I have reached the same feeling about structural and non-structural types. From the language and the rules, it appears non-structural types are defining a domain/scope not residing in the occi core... While saying we are going to treat them like tags, but since they are out of scope ?? how do we interpret them ? I don't think the the occi core is the correct place for adding support for extensions. The process of adding extensions is a different type of use case than modeling occi resources. Extensions requires a different UML diagram to define it. "An OCCI extension model." This unstructured type should be removed from the core model. -gary On 11/8/2010 4:49 AM, Ralf Nyren wrote:
Hi Michael, thanks for your feedback. Very good points you make. Reply below.
On Sun, 07 Nov 2010 05:35:34 +0100, Michael Behrens <michael.behrens@r2ad.com> wrote:
2-cents: Structural and Non-Structural concept might be confusing to folks reading it the first time through. Perhaps its purpose (extensibility) could be stated before their definitions in a non normative manner. Lastly, would adding two subclass of kind (structured/unstructured) help clear things a bit? (The text seems to speak as if there are two subclasses). Perhaps an non-formal intro text in Core would solve this. We could put some more context behind the core model, why it exist etc, without going into the specifics. Any objections?
Regarding structural/non-structural Kind the Kind type does not gain the structural/non-structural property until it is instantiated. The Core doc says in a few places that "a structural Kind" is an _instance_ of Kind etc. However, it can indeed still be quite confusing.
Splitting Kind into two classes each inheriting Category might be a solution. That would automatically solve the issue that a non-structural Kind MUST NOT be related to a structural Kind. I'll see if I can draw up a UML example. Anyone else with an opinion on this?
regards, Ralf _______________________________________________ occi-wg mailing list occi-wg@ogf.org http://www.ogf.org/mailman/listinfo/occi-wg