standard schemas: defining and extending

[Ray Plante]

Hi,

I wanted to share a few thoughts I've had about how I expect 
resource metadata schemas will be defined, standardized, and extended.  I 
don't think these are particularly controversial, but it would be good to 
test them.  (In particular, tell me if you disagree.)  Hopefully this will 
help guide us through some of the items currently under discussion.

I'm motivated by Anita's recent feedback and some of the discussion it 
inspired, in particular, the questions about whether such-n-such should or 
should not be included in the RM document.

So, here's my thinking...
 1. I'm expecting these schemas to be standardized using the process 
    outlined by the DM WG, which can be seen as a superset of the process 
    set forth Document and Standards.   The RM document, a prose document,
    represents the first step.  (According to DMWG, it should include UML 
    diagrams, which we haven't done, yet.)  A specific XML schema based on 
    the document follows, being approved separately.

    (The schema may be in one file or several, as AG's iter. 2 has done.)

 2. The RM document represents only a portion of the metadata that may 
    appear in a registry.  It focuses on:
      a.  metadata generally shared by all resources
      b.  other metadata that may not be strictly universal but which can 
          be used to describe more specific types of resources (such as 
          services). 

 3. The Resource metadata structure defined by the XML schema sets forth 
    the mechanisms for extension--in particular, for describing more 
    specific types of resources.

 4. Other specific types of resources will be defined with their own 
    documents and XML schemas which go through the standardization 
    process separately.  Examples would the extensions for Data 
    Collections, Organizations, Tables, etc.  This allows us to 
    standardize things piecemeal and as needed, rather than having to get 
    it all right at the beginning.

    For services, I expect to see a general extension to cover common, 
    non-standard services, including browser-based web forms, HTTP 
    GET/POST services, and custom Web Services.  In addition, each 
    *standard* service (e.g. SIA, Cone search) will have its own schema, 
    extending the general service metadata.  Again, each of these goes 
    through the standarization process separately.  

 5. The Space-Time metadata is sufficiently complex but self-contained 
    that it should go through the standardization process separately.  
    Although it does not describe a type of resource, it will be used 
    within a resource description; that is, a resource schema will include 
    it.

So as for Anita's additions, some should be included formally into the RM
document; others may be relegated to a schema to be defined separately.  
Regardless, they need to be covered somewhere.  There is no problem with
our prototypes extending the RM (that is certainly the case for the NVO
prototype), and for our January demos, I expect we will need to use at
least some of the same extensions.

Does this make sense?

cheers,
Ray