[tei-council] oversimplification: <measure> isn't measured

[Syd Bauman]

In Berlin the TEI Council requested the creation of a new element,
<measureGrp>, which would replace the special-purpose <dimensions>
element. Was part of the point of this exercise to eliminate the
special-purpose <height>, <width>, and <depth> as well, using (e.g.)
<measure type="height"> or some such instead?

In any case, the implementation as it now stands seems problematic.
Most importantly, the <measure> element has been added to the class
att.measured and removed from att.measurement, which means it has
traded in its quantity= attribute for an extent= attribute. This
seems just wrong. "Extent" has a connotation in English of distance
or area, and is not as general in its meaning as "quantity", which
conveys exactly what the semantics of the attribute are for the
<measure> element. After all, there are lots of things that are
measured for which the semantic "extent" does not apply:
e.g., the number of liters of gasoline (or the number of litres of
petrol :-), the number of cookies in the cookie jar, the number of cords
of wood, the baud rate of a modem.

Secondly, the "suggested values include" list for unit= has lost
all the suggested values that don't make sense when measuring a page.
For the <measure> element, I don't see why TEI should not provide
standard values for all sorts of possible units. There really is no
point in having some encoders use "m", some use "meter", and others
use "metre" to express the same thing. While I admit it is a long
list, there exists an international standard for many of these, why
not use 'em?

Furthermore, the <measureGrp> element permits text content. Is there
a reason for this? I'm not sure it makes sense, but may well be
missing something. No other <*Grp> element permits mixed content
except the dictionary element <gramGrp>. (Although <personGrp> and
<specGrp> do permit <p>.)


Proposal
--------

* move <measure> back to att.measurement

* remove text from content of <measureGrp>? (if it seems important,
  put model.pLike in?)

* change <height>, <width>, and <depth> to a single element <extent>,
  which bears a dir= or dim= attribute whose value may be one of
  "height", "width", and "depth". This new element could be empty
  (quantity always expressed on quantity=, extent=, or value=
  attribute) or could have content of text or macro.xtext. (I'd leave
  it to David and Matthew to decide on that.)

* do this by changing att.measured to att.extent; this class gets the
  new dir= or dim= attribute, has a unit= with only sensible
  suggested values (as it now has -- no change needed) and loses the
  commodity= attribute, which makes no sense. It also gets a
  quantity=, extent=, or value= attribute of data.numeric.


What do you think?

For the old attributes of att.measurement, see
http://www.tei-c.org/release/doc/tei-p5-doc/html/ref-att.measurement.html
For the current attributes of att.measured, see 
http://tei.oucs.ox.ac.uk/Guidelines/en/guidelines-en.xml.ID=REFCLA#att.measured