Registry Interface Working Draft Discussion Topic

Should we add a harvestFrom attribute to the VOResource schema?

Both the STScI registry and Carnivore use an internal atribute called harvestFrom to track where they get records from. This could be useful to expose when investigating problems in the record contents.

There is another way to determine a record's origin based on its authority ID which involves searching for the Registry record that claims management of that ID. However, this is not necessarily the same as where a registry harvested the record from (though it is in practice today). In addition to addressing this ambiguity, a harvestFrom attribute attached to the record itself is easier to get.

Since this piece of metadata does not originate from the resource provider but, rather, really describes the registry that manages the record, it would be encoded as an XML attribute of the Resource element.

If you have an opinion, question, or comment about this topic, feel free to append your discussion below. Be sure to indicate your name as the author.

I am against this idea, as this should be implicit from the AuthorityID, and only one registry should claim to the the owner of a particular of an AuthorityID. In addition I think it should be a principle that a VOresource record should be the same for a particular ID wherever it is retrieved from, and the only place that can make changes to the record is the original publishing registry.

This does rely on making clear in the schema though when a registry "owns" an AuthorityID and when it merely allows you to search an AuthorityID. One way of doing this is to attach an <OwningRegistry> element to the AuthorityID element that points to the owing Registry entry, and then in the Registry entry have a series of <SearchableAuthority> element for the authorityIDs that it has harvested.

-- PaulHarrison- 07 Apr 2005