[oslc-cm] Concern on property names with both purpose and embedded resource type information

[Steve K Speicher]

In CM 2.0 there are a number of property names that include both the 
meaning of the property (implements) and the expected type of the 
target/object (Requirement), you we have things like 
oslc_cm:implementsRequirement [1].  This has provided some usefulness in 
being able to know from a client perspective what to expect.  This has 
also limited what a client thinks it can put or expect as the target of 
that link.  What had motivated the previous design may not be best 
practice anymore.  So I'd like to use this email to make people aware of 
this topic and gather some feedback.  Also I'd like to discuss strategies 
to introduce new predicates, such as oslc_cm:implements and how to 
potentially phase out or deprecate things like 
oslc_cm:implementsRequirement  Another example of simplification would be 
to replace all oslc_cm:related* with dcterms:relation.  We can produce 
some guidance and explain how tools can efficiently know what type of 
resource is the target of the link without having to encode into the 
property name.

This is not a 2.0 issue, suggesting this as a topic for how we evolve the 
vocabulary in 3.0.  I've heard this issue from a number of adopters who 
needed to invent new properties that were "like" currently defined ones, 
it would be good if they weighed in as well.

[1] - 
http://open-services.net/bin/view/Main/CmSpecificationV2#Resource_ChangeRequest

Thanks,
Steve Speicher
IBM Rational Software
OSLC - Lifecycle integration inspired by the web -> 
http://open-services.net