Registry Interface Working Draft Discussion Topic

Should we make the harvest() function independent of the searchable interface?

Section 3.2 of the RI describes a currently unimplemented function that allows harvestable registries to tell harvesters to start harvesting. This is the only way a harvester can learn about a new harvestable registry (apart from someone sending an email). This question concerns whether this function should be strictly tied to the searchable interface, or should be made independent. This affects how the RI WSDL is defined.

In the latter case, one might have a resource that harvests for a reason other than to populate a searchable registry. An example of this is the NASA ADS who is interested in integrating VO resource records into its abstract database. (See also the GetRegistries discussion topic.)

Note that currently the harvesting interface is independent of the searchable interface as a registry can operate as either type.

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 think we should have an independent harvester interface, but indicate that searchable registries SHOULD implement the harvester interface as well. If we keep/redefine getRegistries() (see GetRegistries discussion), we won't be talking about just on function in the harvester interface.

-- RayPlante - 30 Mar 2005

Resolved: The decision to set up an IVOA-sponsored registry of registries makes the harvest() function superfluous, and will subsequently be removed from the standard. No general harvester interface is needed.

-- RayPlante - 05 May 2005