how to make a publisherDID

[Douglas Tody]

On Wed, 26 Oct 2011, Patrick Dowler wrote:

> On 2011-10-26 07:33:08 Douglas Tody wrote:
>> So far as conventions go my preference is still for something like
>>
>>      ivo://cadc.nrc.ca/archive/cfht#12345/raw
>>
>> where the convention is that ivo://<authority-id>/<collection-name>
>> (everything before the #) is also the collection identifier and the
>> remainder is the specific data product ID within the collection.
>> Perhaps this should eventually be more than just a convention.
>
> Yeah, I did this because the stuff before the # is in fact a regsistered
> DataCollection, but according to the discussions this means that there is one
> identified resource and that in principle one could resolve it, get it, and find
> the fragment within. That is definiotely *not* what we want to say.

I don't understand this part - it is not how I thought this works.  The DID
points to a single data product.  The authority ID is still there and unchanged
in all these examples.  There is nothing formal about collection in a DID
so no implication regarding what one can do with the collection.

> I can live with <authority>/<dataset-specific_id>, but if we want to enable
> someone with a publisherDID to find and download that dataset it presupposes
> that the client could find services owned/operated by that authority and invoke
> the service(s) with the publisherDID.

Yes - some extra registry capability may be needed to enable this.  Or possibly
a global indexing service that integrates and indexes all the data products.
Then one could just search by the pubDID and the data links (if provided) would
point to the relevant services.

> Alternatively, if it means (in future)
> they look for a specific type of service under that authority then it is more
> or less manageable, but having multiple services is still problematic. I
> personally don't care much about that because we are trying very hard to
> provide common services across all our collections, but other people may not
> like it.
>
> I realise that publisherDIDs seem quite disconnected from using services, but
> I can't ignore this idea that being able to resolve them in a prescribed way
> seems like a really good idea.

We don't currently event require services like SIA/SSA etc. to provide search
by pubDID but ultimately we will need to, once this is more broadly implemented.

 	- Doug

> -- 
>
> 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
>