Registry discussion

[Tony Linde]

Hi Roy,

Your last point is taking you around in circles. If the registry that is
queried for resources does not have all the resources (or access to a 'full'
registry) then how is it going to find the roy.org resource in the first
place? And if it already 'knows' about it, why bother with a registry?

Cheers,
Tony. 

> -----Original Message-----
> From: Roy Williams [mailto:roy at cacr.caltech.edu] 
> Sent: 15 September 2003 16:23
> To: ael at star.le.ac.uk; registry at ivoa.net
> Subject: Re: Registry discussion
> 
> 
> > I thought we had something relatively simple but still 
> extensible with 
> > RM v0.8 and schema v0.8.1.
> 
> Agreed. The schema is fine.
> 
> > > ** Compatibility with ADEC is the most important new thing in the 
> > > registry discussion. We should concentrate on that.
> >
> > This has popped up in the last week and suddenly it is the most 
> > iumportant thing in the VO world. Why?
> 
> The reason we do research is to get journal publications. 
> Once you have published to VO, it would be good if journal 
> publication (ADEC) is straightforward. I would like for the 
> systems to be joined -- that way the journals would cite VO 
> identifiers. I would like for publication to happen in 
> stages: a personal digital library, then sharing with a 
> group, then public, then peer-reviewed.
> 
> > > ** I don't like the idea of "global registries" or "full 
> registries" 
> > > that are mandated to hold everything in the VO.
> >
> > Again, why? If we come up with harvesting/replicating protocols why 
> > won't
> it
> > work? I cannot believe that there'll be that many resources.
> 
> Its OK for registries to make an effort at replication. But 
> they cannot promise to have everything. There is the time-lag 
> for add/delete does the submitter get an email when their 
> resource is "really" in the VO? There will need to be 
> rejection criteria for maniacs and nutters. There will be 
> people like me who think that all 10000 Vizier catalogs 
> should be registered or else the thing won't get used 
> (maniac? nutter?).
> 
> > <<people can choose what their IDs look like. I think the ID should
> reflect
> > where you can go to resolve it. If the ID says roy.org, then ...>>>
> >
> > That sounds more complicated to me. Instead of hitting one registry 
> > for
> your
> > metadata, you end up chasing all over the shop hoping that all the
> relevant
> > services are running.
> 
> If there is just one registry, who owns it? If more than one, 
> how many? If unlimited number, then we are chasing all around 
> the shop. Solution: the ID points to the place that has the metadata.
>