how to make a publisherDID

[Douglas Tody]

But this is a query string and not a fixed identifier of the sort we
want to use to index archives; the pubDID will often be used as an
actual key to build DBMS indexes (the proposed syntax might work
technically as a key but is not in the spirit of what is intended for a
DID).  Also this contains reserved characters which are not currently
legal in an IVOA dataset identifier ('?', '&', probably '=').
Everything which is legal in URI is not necessarily legal in an IVOA
dataset identifier.

I don't see any reason to change the simple nature of a DID which is an
authority ID concatenated with a (unspecified format) resource key.  We
just need to use this as a simple key as intended to go look up
additional information stored elsewhere.  The pubDID can be used to find
the ObsCore record for a data product which contains detailed
information in a standardized form.  The authority ID can be parsed off
and used to access a registry record which can be extended if we need to
store additional information at that level.

 	- Doug

On Wed, 26 Oct 2011, Mark Taylor wrote:

> Pat,
>
> URI permits a query part, so one possible way would be something like:
>
>  ivo://cadc.nrc.ca/archive/cfht?observation=12345&product=raw
>
> As far as I can see this would in principle solve the parsability
> issues you're looking at, and it's more flexible than putting
> everything in one hierarchical path.  It's probably a good enough
> solution for human users.  However, given that it's not
> currently standardised, the answer to how a machine holding one
> of these actually works out what it means [still] relies on
> it's understanding your particular conventions for encoding
> the information in a URI.  Whether a human and/or machine actually
> *needs* to understand what it means is another question, I'm not
> involved enough with this stuff to have an opinion.
>
> Mark
>
> On Tue, 25 Oct 2011, Patrick Dowler wrote:
>
>> I need to decide on a procedure for making publisherDID values for use in TAP
>> services, for example in the ivoa.ObsCore table, but also our internal CAOM
>> tables which also expose the same identifiers. This is following from the
>> interesting DAL-DM-DCP discussions on use of URIs.
>>
>> Presumably, I could start with a DataCollection identifier registered in the
>> IVOA registry system, eg:
>>
>> ivo://cadc.nrc.ca/archive/cfht
>>
>> (note: more generally, a creatorDID would be linked to a DataCollection and a
>> publisherDID might be linked to a provider/service in case they are not the
>> same or definitive provider - let's ignore that for now)
>>
>> In the past we would use the fragment, eg
>> ivo://cadc.nrc.ca/archive/cfht#12345/raw but that implies that one can somehow
>> "get" the entire collection and then look for "12345/raw" inside it (client
>> side).
>>
>> Instead, I could append some dataset-specific to get an observation identifier,
>> eg.
>>
>> ivo://cadc.nrc.ca/archive/cfht/12345
>>
>> and then append something product-specific to get an identifier for the product,
>> eg.
>>
>> ivo://cadc.nrc.ca/archive/cfht/12345/raw
>>
>> However, if someone has one of these (the last one), what can they do with it?
>> With this approach, there is no prescribed way to extract the identifier for
>> the registered DataCollection. Should there be?
>>
>> In VOSpace, the spec invents a new URI scheme (vos) that prescribes exactly
>> how to extract/create the service URI from the vos URI, but it comes down to
>> being able to separate the URI into two halves: a base that can be looked up
>> in a registry and a separate bit (path in vos) that can be used when talking
>> to the service. In principle this does not require a new scheme, but it does
>> require at least a way to separate.
>>
>> PS-I have the exact same issue with several services where I want to identify
>> the service with an ivo URI and I want to make and use URIs for items found
>> within/via that service.
>>
>> --
>>
>> Patrick Dowler
>> Tel/Tél: (250) 363-0044
>> Canadian Astronomy Data Centre
>> National Research Council Canada
>> 5071 West Saanich Road
>> Victoria, BC V9E 2M7
>>
>> Centre canadien de donnees astronomiques
>> Conseil national de recherches Canada
>> 5071, chemin West Saanich
>> Victoria (C.-B.) V9E 2M7
>>
>
> --
> Mark Taylor   Astronomical Programmer   Physics, Bristol University, UK
> m.b.taylor at bris.ac.uk +44-117-928-8776 http://www.star.bris.ac.uk/~mbt/