Re: [dxwg] Should a server list all profiles it supports, even those in a hierarchy (#932)

+1 @smrgeoinfo @nicholascar 
Whatever definitions we add, requiring clients to use the "hierarchy of profiles" is going to be very demanding, as they would have to follow their nose to the profile description and then do some inferencing over the links they find there. 
I think enabling such thing to happen is good, especially in case the data services are "lazy" (i.e. for some good or bad reason or another they only publish a subset of the profiles they conform to, see note) but we shouldn't require it. So servers SHOULD list all profiles they support.

Note: there's a limit to our recommendations though, and this where the "inferencing" would be useful: first, the designers of a service may not be so much aware of profile derivations for the specification they use. Then it could be that designers of different agents don't have the same perception of what qualifies as a "relevant" base for a profile. I.e. a DCAT-AP server could count only DCAT as a valid base profile, and thus advertise these two profiles. But others (and we had this discussion) would consider Dublin Core to be a relevant base profile too, as DCAT re-uses some of it. We won't solve that I believe, so we have to engineer around it. And with I believe so far we're quite in a good position for this, with that PROF enables us to do.

GitHub Notification of comment by aisaac
Please view or discuss this issue at using your GitHub account

Received on Tuesday, 6 August 2019 14:06:21 UTC