|
I do not understand why there need be an interface for the delta deployments. Well, it may be beneficial to have such a method somewhere, but:
I suggest we spare the effort. I understand Stefan's idea.
I think it would be an effort worth it because: As I understand Stefan the delta-deployment-interface is intended to be an extension to the (I say) root-deployment-interface. With these methods we provide a somewhat generic way for deployment: The root-deployment methods support deploying and undeploying on the root level and the delta-deployment methods support any kind of change to an existing policy tree. The latter is also implemented very generic due to the fact that the delta-objects are of type DeploymentModification that is only a marker interface. In my opinion this approach suffices a lot of "standard" cases and if not the user is still open to use his own, custom, interfaces. I agree with Stefan with respect to Wolfgangs input (they both are right).
My opinion is that we should merge both ideas:
Interpretation:
Additionally implemented some small corrections from the Review HERAS-AF-XACML-14.
And also renamed the policy repository interfaces to the names, which I discussed with Wolfgang.
PolicyRepositoryDeployment --> PolicyRepository - PolicyRepositoryEvaluation --> PolicyRetrievalPoint |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
One interface to get the policies (during evaluation), one Interface to deploy undeploy root policies and another one to deploy deltas to an existing policy tree. I don't think that everyone who wants to implement a repository wants be able to implement the possibility to change an existing policy tree.