On May 16, 2017, at 5:48 PM, Stefan Paetow <Stefan.Paetow@JISC.AC.UK> wrote:
Ok,
instance. Do I simply name them (i.e. name_of_query = 'query here...') and then just invoke them in their appropriate section?
It depends on what you want to do. What's the schema and use-case?
Schema is something like this:
[gss_acceptor] [namespace] [username] [targeted_id]
The *-TargetedId entries depend on 3 attributes: GSS-Acceptor-Host-Name (gives us Moonshot-Host-TargetedId), GSS-Acceptor-Realm-Name (gives us Moonshot-Realm-TargetedId) and Trust-Router-COI (gives us Moonshot-TR-COI-TargetedId).
Then you're probably better off just writing a custom SQL inner in the post-auth stage...
Given the policy unlang to generate a targeted Id is exactly the same for all three versions, except for the namespace and the GSS attribute, either I define custom attributes in the UKERNA dictionary that I set the gas_acceptor, namespace and targeted_id to for each generation, or I change the schema to include all three like this:
[gss_acceptor_host_name] [gss_acceptor_realm_name] [trust_router_coi] [namespace] [username] [host_targeted_id] [realm_targeted_id] [tr_coi_targeted_id]
This just seems a waste of space.
What do you think is better? The first option? It seems more concise.
I agree. Alan DeKok.