ACL Expiry about web-access-control-spec HOT 6 OPEN

i would let the apps clean the acl (so acl can be more transparent and policies would be easy to renew)

from web-access-control-spec.

Note that RDF is monotonic, so if a graph is true, deleting statements leaves it true. Having an optional constraint like acl:validTill 2016-06-30 doesn't work like this. Ways to get around it are many -- have a separate class TimedAuthenticaion for example where the validTill is required. This also means old systems will fail safe.

from web-access-control-spec.

@timbl Can you explain a bit more, about monotonic? What do you mean about deleting statements? Does that mean our PATCH verb is useless?

from web-access-control-spec.

@nicola I agree the cleaning process shouldn't be part of the specs, but I believe the server is more in charge of cleaning. It can happen with every write to the resource or it can rely on server implementation. Poor applications shouldn't cause a lot of dummy triples on the pod.

@timbl does this mean in an authorization rule containing acl:validTill, removal of acl:validTill triple should leave the rest of the triples in the same authorization valid which is not the case here. Removal of acl:validTill triple invalidates the whole authorization rule. I hope I got it right.

In this case, why don't we consider Authorization is actually a TimedAuthorization which has acl:validTill mandatory and for unlimited authorizations, the object could be blank or a very far time in the future? does this make sense?

from web-access-control-spec.

@dmitrizagidulin do you think @timbl 's point also applies for acl:defaultForNew as it is also optional?

from web-access-control-spec.

@ghanemabdo not sure.. I need to understand this subject better; I'll see if I can ask him about it today.

@nicola - don't forget that many (most?) apps won't have read/write access to the .acl resource, and so won't be able to perform cleaning.

from web-access-control-spec.