content, format, ctype, or xtype ?

[Markus Demleitner]

On Tue, May 12, 2009 at 02:36:52PM +0100, Mark Taylor wrote:
> On Tue, 12 May 2009, Paul Harrison wrote:
> > Can this concept not be satisfied by the units metadata for a column? There is
> > a "units" effort going on within IVOA at the moment - if this prescribed the
> > format for certain specialized units like MJD, sexagesimal etc. then there
> > would be no need to add a piece of "format" metadata
> I'd be happier with this solution than with the status quo
> *if it was written into the standard rather than just used in an
> ad hoc basis as now*.
Though I'd be still much happier if there could be a separation of
concerns like my understanding so far:

(a) ucd: What's in the column?
(b) unit: What's the unit of the values?
(c) utype: What's the role of this column in a defined data model?

plus, I'd wish,

(d) content, ctype, representation, whatever: What representation for a
    value has been chosen?

In particular, I feel mashing together (c) and (d) is unattractive
since (d) is necessary even for data not conforming to a specific
data model.  

I agree with Mark that mashing together (b) and (d) would be
preferable to this if ctype-like attributes are out.  But it's still
clearly overloading two concepts into one attribute.  That's is not
necessarily bad, but in this particular case, I'd have doubts such
overloading would help either implementors, users or the clarity of
the specifications.

There's one thing going for stuffing ctype into units, though: What's
the unit of the components in a construct like

<FIELD unit="deg" ctype="stc-s"/>

with a value of

Point FK4 1.1 -1.2 unit rad rad

Or, worse,

<FIELD unit="rad" ctype="stc-s"/>

with a value of

Point FK4 1.1 -1.2

(STC-S has a default unit of deg in this case).  Or even:

<FIELD unit="deg" ctype="stc-s"/>

with a value of

Time 2009-05-12T15-13-00 unit s Error 6

So -- I'd feel that ctype="stc-s" would preclude the use of unit.
Arguably, a similar point could be made for non-STC time
specifications.

That would *almost* convince me it's a good idea to merge unit and
ctype...


Cheers,

       Markus