Comments on the new Schemas

[Ray Plante]

Hi Aurelien,

Thanks for the updates--very helpful and detailed.  I've posted corrected
versions to RegUpgradeSummer2006 (and to VOResourceV10).  Here's what I
did:

On Fri, 16 Jun 2006, Aurelien Stebe wrote:
> - The "validatedBy" attribute on ValidationLevel should be made 
> "required", because a level number without any indication about who 
> validated it is not very useful ... or do we assume it is the 
> queried/managing registry if empty ?

made it required as suggested.

> - The "managedAuthority" element should have minOccurs="1", since a 
> registry must manage at least its own AuthorityID.

As mentioned before, registry could be a searchable only and therefore 
would not have any managedAuthority elements.  no change.

> - The "validationLevel" is missing in "RegCapRestriction" (VORegistry.xsd)

fixed.

> - We should rename OAIHTTPGet to just OAIHTTP, ...

done.

> - In the Registry extension, the first documentation paragraph should 
> say : A registry is considered a publishing registry if it contains a 
> capability element with xsi:type="vg:Harvest". It is considered a 
> searchable registry if it contains a capability element with 
> xsi:type="vg:Search".

done.

> - StandardSTC might better be declared in the VOStandard schema, but I'm 
> not sure I understand how to use it. Also, I didn't fully read the 
> VOStandard file as I think we should concentrate on this schema after 
> the RI and the basic ones are in v1.0

As mentioned before, StandardSTC is needed for our use of coverage, 
defined in VODataService, so it remains there for now.

Your note made me look more closely at this, and I discovered that I had 
not integrated the lastest changes I discussed with Arnold.  I did that; 
in particular, I updated the stc.xml example to illustrate its use.  

> - Inconsistencies with "whiteSpace collapse" : it is missing on some 
> elements. Here are a few examples : it should be on "AccessURL" element 
> and not on its attribute "use"   /   "wsdlURL" should be of type 
> "PaddedURI"   /   "AuthorityID is missing the "whiteSpace collapse"   
> /   missing in the whole SIA and CS schemas.

Yes, you are right.  This made me look over all my use of whiteSpace as 
well as PaddedString and PaddedURI.  I discovered that I had 
misinterpreted the definitions of some of the standard types--particularly 
xs:token and xs:anyURI.  That is, it turns out that the Padded* types are 
not necessary as they are equivalent to the xs:token and xs:anyURI.  I've 
dropped the Padded* types and replaced their use with the corresponding 
standard types.  

> - I noticed that the only difference between DataService, CatalogService 
> and TableService are the Coverage and Table elements. Those two elements 
> are optional and I think we could put them both in DataService and get 
> rid of the two other extensions. Those three types are just confusing 
> users. A service for which coverage is irrelevant would simply not have 
> this element, and one without "tables" would not declare any.

As mentioned before, the semantic meanings of these types have use by
themselves.  The recent name change was meant to reduce user confusion.  
That is not to say that no confusion remains.  These types were created
(as a compromise) because some people did not like allowing the metadata
mentioned to be attached to resources to which they did not make sense.
Under the assumption that this argument is still relevent, I'm inclined to 
keep these types.  

thanks!
Ray