Consider normalizing identification responses based on Cat-Interop work

[roomthily]

See https://github.com/OSGeo/Cat-Interop related to ESIP best practices discussion.

(See also the OSGeo geolinks but it assumes "correct" OGC urls and that is not always the greatest.)

[roomthily]

Forked the Cat-Interop repo to the bcube org. Going to use the identifier field for our protocol, service, dataset, etc, identification.

[roomthily]

We will need to expand that list quite a bit.

[roomthily]

TASK: Revise the list for all identified services (I don't expect some/many of them to be of interest to the osgeo group but maybe add a field to flag those?)

[roomthily]

Feeling like I should standardize their identifier structure as well:

WWW:SPARQL:1.1 vs OpenSearch1.1:Description

and add some flag for actual service instead of some of the more generic terms (order, download - these are reading like codelist values and are being conflated here). There's service identification and how to map those into ISO 19115 (that is what they're doing) and there's data access identification which is just a descriptor of that service (or part of that service). (Seriously, they are blending data formats with service descriptions and all kinds of things in a really awkward way.) (missed the resource_type field, my bad. but still not complete.)

But, the identifier structure. So some namespace-like term (WWW or OGC or whatever), service type/name (SPARQL, WCS, etc), version, response type (catalog vs dataset vs ?? this is a tricky bit). The two examples do not match structures.

[roomthily]

And also some thinking over there about URNs:

OSGeo/Cat-Interop#3

so check on that and also the resolver concerns:

http://lists.osgeo.org/pipermail/cat-interop/2014-March/000027.html