Re: FYI - Regrep Reference Implementation - Questions/Issues

[Terry Allen <tallen@sonic.net>]

Laura wrote:
| These are the notes I took when Una was reviewing her slides ....

Thanks for this, Laura.

| Policy Questions/Issues

I agree these are indeed mostly policy considerations.  My
responses here are opinions on policy.

| What can people submit?
|  - XDR?

Anything relevant to the purpose of OASIS?  I don't see how we
can exclude any application of SGML (and I see XML and its
various schema flavors as applications of SGML).

| Shall submissions be uniform and if yes, what should they include?

I don't think we'll get much response if we demand uniformity.

|  - schema
|  - sample instance
|  - supporting documents

Just a bare DTD would be acceptable to me, although of course
it needs its accompanying metadata.  We might want to require a
short description.  Remember, one of our goals is supporting SGML
on the Web (download of DTD for parsing an instance).

| Is every submission reviewed before being accessible on the repository?
|  - Assumption is yes

Agreed.  However, in the EBXML work, there is a requirement that
trusted users be allowed to deposit without review, or with review
only by software check that the deposited item matches "the 
metamodel".  I think we might permit at least some users to be
trusted.

| Are updates allowed?

They must be.  But I think this question is about the distinctions
between "update" and "version" and "new submission".  We need some
use cases here to organize our discussion.

| One submitter per organization?

See my message of Saturday.

| What should be available for System Download?
| (what is diff between download and system download)

Yes, I don't understand the distinction.  Every registered item
must be downloadable, and the metadata (registry information)
about it (perhaps with certain contact info deleted) must also
be downloadable.  

| Site & Maintenance Issues
| 
| XML.ORG Site look & feel
| Admin and SME - Staff
| Min Browser Support
| Testing Support - Q&A
| Prototype and Timeline
| Licenses (Oracle)
| XML Interfaces 

means what?

| Regrep conformance

The conformance language is still to be written (or collected; some
of it is sprinkled throughout the tech spec).  NIST will help on
this.  I've proposed that we not try for consistency in method of
submission; if we agree on that there would be no conformance 
requirement for submission.  We *must* conform to RFC 2483 as
unofficially revised, so as to present a uniform method of access.

| Need Sample Data

Yes.  Can our managing editor arrange for this?

regards, Terry