roadmap

[Markus Dolensky]

... all understood. Yet again, the question remains:

At which point in the roadmap do you foresee/suggest the inevitable jump into
the cold water?
> Which bullets (1-12) in the roadmap need to be done to reach this point?

2004, 2005, ... ?

Markus




Tony Linde wrote:
>>maturity of the underlying Schema is secondary to content 
> 
> 
> I'm not sure the underlying schema is secondary. I *wish* it was. It would
> be good to say, let's forget the schema - any registry builder can do what
> they want - all we specify is the interfaces that the registry has to
> support.
> 
> But for the interfaces to work, information has to be exchanged between
> registries (harvesting), between registry and dataset (publishing) and
> between registry and application (query). And it is the format of that
> information that we have to specify.
> 
> An application has to know what type of query it can make and what type of
> information it will get back. If the query is of the type 'which tools can,
> from a photometric catalogue, find the redshift of each object by means of a
> standard SED fitting procedure' or 'which data services contain provide
> catalogue data about white dwarfs in patch X of the sky' or ... then it
> entails that the registry supports metadata about resources of the type
> required. And that is why we need the schema to be sorted out in agonising
> detail - unfortunately.
> 
> Let's face it, putting entries into a registry is pointless unless they can
> be used. And if all we use the registry for is to list datasets and their
> web address then we might as well just list them all on a web page. The
> metadata associated with datasets, tools etc must be rich enough to allow
> *applications* to do real work with them. (And I'm not saying all the
> metadata has to be in the registry as AstroGrid is doing - the metadata can
> be held at the resource - but it must be accessible from the registry so
> that registry queries can be *completed* in a single call.) Making the
> resource metadata too simple is worse than useless since it makes the whole
> VObs effort look like nothing more than collating a list of datasets.
> 
> Cheers,
> Tony.