[OAI-implementers] Open Archives Initiative Protocol for Metadata Harvesting Version 2 news

Simeon Warner simeon@cs.cornell.edu
Mon, 4 Feb 2002 23:22:37 -0500 (EST)

On Mon, 4 Feb 2002, Josie Imlay wrote:

> I was thinking that on the sets being optional it would make more sence
> if a particular archive does not want to use sets for their to be a
> standard responce, or way to handle this.

The standard way to say 'I do not implement sets' is a to list no
sets in the ListSets response.

> To me the way to do this is to make sets manditory, but say that if you
> don't want to use them then just have one set called "all" or something.
> This is hardly makes it any more dificult from the standpoint of the
> developer because he can just have a static responce. But from the
> harvester that's programmed to deal with sets, it doesn't even notice
> that there is a difference.

no setSpec => universal set => all records

Consider two cases where a harvester might want to use sets:

1) to harvest part of a repository. This requires prior (outside the
protocol) knowledge of the set structure and hence the set or sets to be
harvested cannot be determined automatically.

2) to mirror/harvest the whole repository including the set structure. In
v1.1 it is necessary for a harvester to harvest from all sets and from the
universal set (no setSpec) to be able to deduce the set structure. In v2
all records will include set membership information so it will be
only be necessary to harvest the universal set.