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

Simeon Warner simeon@cs.cornell.edu
Mon, 4 Feb 2002 15:34:10 -0500 (EST)

I have always failed to see why sets cause so much confusion. Sets ARE 
(and will remain) optional! They are optional from both perspectives:

1) a repository may choose not to implement sets (only overhead is
providing a blank ListSets reply)

2) a harvester may ignore sets by never calling ListSets and never
specifying a 'set=' parameter.
I think there is sufficient interest in sets to keep them for those that 
do choose to use them. 


On Mon, 4 Feb 2002, ePrints Support wrote:
> I strongly feel that sets should be made optional (maybe flagging this in
> the Identify response) to reduce the implementation load on small archives.
> Sets do not appear to be widely used by harvesters (shout if this is wrong)
> and increase implementation complexity by about 25%. The "which sets is this
> in" query makes this even more fiddley.
> It's easy for harvestors not to support sets, but if it's meaningless which
> it seems to be for some (many?) archives then forcing them to do extra work
> reduces the number who will uptake.
> also:
> > 10. Multiple metadata formats - Modify ListIdentifiers to permit a metadata
> > format as argument, filtering the return to include only record identifiers
> > that support the specified format.
> This is also more complexity, why is it needed?