Sets for Registry Interface

[Gretchen Greene]

Hi Kevin,

Glad to see the doc is coming along.  

Hopefully this doesn't sound too nit picky here,  but in rereading the
section on sets I think there can be some clarification made because
these are going to be more key to harvesting uniformly. Is this TBD with
Appendix A.3?  Specifically, the first sentence quotes from the OAI ref
'optional construct' then at the end of the paragraph there are two
required sets for the harvestable registries.  While those of us that
are trying to adopt the RI standard are familiar,  it is a little
confusing if you aren't reading the full explanation.  One might skip
over the section after reading 'optional' (in fact I started to speed
past this in deferrment but realized it is more pertinent).

Another point is that since we should support the ivo standards with the
set names,  it would be good to list them out (perhaps in A.3?) to be
clear for the schema versions that have been accepted as registry
'standard' extensions.  If they are not spelled out then there is room
for variation on naming...  Also assuming these will be grouped for all
previous versions supported and not get too messy here.

For example, 

Set names		

ivo_managed		required
ivo_standard	required
ivo_registry	optional
ivo_siap		optional
ivo_cone?		optional
And so on...