A suggested revision for UCDs

[Thomas McGlynn]

A few minutes ago I uploaded a version of my suggested revised
proposal for UCDs to the Twiki.  This is just a Word version since
I don't have a PDF generator handy.  The URL is
    http://www.ivoa.net/internal/IVOA/IvoaUCD/UCD-1.9.9b.doc

This builds upon Roy and Sebastien's division of UCDs into concepts
and propeties but puts them together rather differently.

In addition and largely independently, it includes a discussion
of the use of UCDs in the context of groups of columns and
tables as a whole.  With the approach suggested, I hope that the
ambiguities that have heretofore precluded considering using UCDs
to mediate between data models and data can be removed and we may
no longer require a utype parameter.

Some discussion of the document is included (in red) within the
text.  This mostly describes the relationship of this version to
the 1.9.9 version.

Sections 3 and 6 of the document are copied from the previous version.
(though section 6 was section 8 in that document) from the
previous version.  The abstract is altered to reflect the discussion
of grouping constructs.  Section 2 -- describing the status of the document --
has been changed to discuss the implications of the adoption of this
recommendation upon existing software systems and protocols.  I think
that some such statement should be part of draft recommendations.  Not sure if
that's part of the recommendation process but it probably should be.

Section 4 is the discussion of UCDs and UCD syntax.  At the end of section 4 there
is a long parenthetical discussion of some of the ways this proposal differs from 1.9.9
since people seemed to be concerned about that last Thursday.  If they've
gotten this far they probably don't need this anymore but it may be of interest.
This section tries to be rather rigorous -- addressing a fair number of nits that
had not been talked about in the earlier proposal even though it makes
the proposal somewhat longer (e.g., the discussion of array valued cells,
a much more detailed discussion of comparability of columns, how to ensure the
uniqueness of UCDs, ...)

The actual definition of all of the valid words for UCDs is deferred
as it was in the earlier version, but a substantial number of examples are given.
[In fact, most of the words should transfer between the two versions fairly
transparently.  The differences lie mostly in how they are put together.]

Section 5 is the dicussion of UCDs and grouping structures.  I'm quite
excited by this since I think it real potential for helping to unify discussions
of data, data models and data access.  The discussion
in this chapter is less rigorous -- even if the basic idea is adopted I'm sure
it will need substantially more work but I think it has real possibilities for
linking data models and data.  This chapter is why I've sent this message
to the DAL and DM groups as well as to the UCD group.  Apologies to all of
you who get this twice or thrice!

I trust there will be comments...


	Tom