URLs?

[Alasdair Allan]

Roy Williams wrote:
> Ray Plante wrote:
>> Alasdair Allan wrote:
>>> Ray Plante wrote:
>>>> The previous proposal for the ID had another added benefit.  It 
>>>> recommended that the part before the # refer to the event registry 
>>>> service that is capable of resolving the identifier to the actual 
>>>> event message.
>>>
>>> Err, no. This was never the case IMHO...
>>
>> I just looked more closely at the version I read last week, and indeed
>> what I said is not quite correct, but close....
>
> My understanding is exactly what you wrote. It means that when 
> identifiers are chosen, the part before the # corresponds to a 
> database, and the part after is the sequence number in that database.
>
> However, some of the potential customers of VOEvent want to send out 
> events without the need for archiving them in a database, so they will 
> not be resolvable in the way we thought.  We concluded at the workshop 
> that there is nothing wrong with that. However, I am not sure how to 
> build the identifier in this case.

We concluded during the mini-interop, following the meeting, that while 
the original publisher might not be archiving the original message 
there may be one (or more) distributed repositories which is doing so. 
To retrieve  a specific message at that point you'd go to the Regsitry 
and request a list of repositories that archived a certain publisher. 
You'd then pick one, and retrieve the message from them. It would be 
the responsibility of the repositories to register the list of 
publishers that they were archiving with the VO Registry.

Of course there is nothing to stop a publisher also acting as a 
repository.

I think we've moved to a much more decentralised network, and in my 
opinion this is a good thing! Rob was taking notes during the 
mini-interop and I think we should wait for him to push these to the 
group before moving forward with this discussion to bring everyone up 
to speed with what we discussed there (we got through a lot of stuff).

Cheers,
Al.