Hi all,
Okay, since Clive asked for it....This is not actually a summary/recap of the identifier issues; however, I hope this will provide a bit more clarity to the discussion. Thanks to everyone who was able to contribute to Friday's discussion about identifiers and related issues. I know it had a lot of threads and was often difficult to keep up with; nevertheless, it was very valuable (in my mind, anyway) in helping to understand the current issues of identifier control, relationships to registries, persistance, and location-independence. I mulled over the disussion over the weekend, and I'd like to propose 4, somewhat independent amendments to the Identifier framework.
I'm posting the four different amemndment ideas on separate twiki pages to save you from having to read them all. You can focus on the ones that concern you the most. Here's a summary:
Part A. Establishing AuthorityID Ownership and Resolving
ResourceIDs
http://www.ivoa.net/twiki/bin/view/IVOA/IdAmendPtA
A clarification of what's currently in the Identifier (Working Draft), the main differences beeing: o A request for an AuthorityID is handled through a publishing
registry to global registry. The request is accompanied by the
identifier of the publishing registry to be saved in the global
registry as the "registry of origin" for that authority ID.
o Registries are registered as first-class resources. Their
descriptions should include the AuthorityIDs that they have
resources for.
Part B: Unregistered Resource IDs
http://www.ivoa.net/twiki/bin/view/IVOA/IdAmendPtB
This amendment allows one to associated identifiers with resources that are not strictly registered.
Part C: Support for Persistance and Location-independence:
Persistant Logical Identifiers (PLI).
http://www.ivoa.net/twiki/bin/view/IVOA/IdAmendPtC
This proposal for handling location-independent identifiers is much the same as the strawman I presented to this list earlier (http://www.ivoa.net/forum/registry/0454.htm). It involves use of metadata to describe mirroring relationships, as well as proposing the use of Persistant Logical Identifiers for refering to resources in a location-independent way.
Part D: Persistant Dataset Identifiers (PDI) and ADEC
Compatibility.
http://www.ivoa.net/twiki/bin/view/IVOA/IdAmendPtD
This proposes dataset identifiers based on the PLIs discussed in Part C; these can be used as ADEC identifiers for publishing in the journals. From a PDI, one can use VO registries to discover data resolvers that resolve it into one or more locations to retrieve it from.
Although I'm calling these amendments, not all of it would go into the Identifier WD per say. Some of this is input to registry services.
cheers,
Ray
Received on 2003-09-15Z11:29:34