[tei-council] to-do list & other changes w/ problems raised

Wed Jun 28 10:02:23 EDT 2006

I believe I have completed all of my personal assigned action items
from Kyoto with the exception of "find out places were an element is
place like and datalike", which requires more serious sit-down time
than I will have until after DH in Paris. 

I have also performed a few other little fix-ups to P5, some of which
raise some issues.

* Fixed a bunch of n= attrs of chapters (and one <head>)

* Fixed start= of <schemaSpec> to data.names

* Problem: prefix= of <schemaSpec> should be constrained to an
  xsd:NCName, as that's what RelaxNG expects[1]. We don't have a
  datatype that represents xsd:NCName. Possible solutions:

  1. use data.name (it's awfully close) and mention in the 'remarks'
     section that using a colon will foul up your schema -- this is
     what I've done for now

  2. use data.name and add a Schematron check for a colon

  3. create a new datatype, data.ncname that boils down to
     xsd:NCName, and use it

  4. just use xsd:NCName directly (running afoul of our datatype
     system)

* Made extent= of <space> data.words. (It was 'data.enumerated',
  which seems ridiculous.) This is not an ideal solution. It would be
  better to have quantity= and unit= from att.measurement, but the
  problem is we don't want commodity=. Possible solutions:

  1. Use att.measurement and live with a ridiculous commodity=
     attribute on <space>

  2. Just copy the quantity= and unit= attributes over from
     att.measurement to <space>. This has the enormous advantage that
     the suggested values include list of unit= can list values that
     make sense for <space>.

  3. Divide att.measurement into two attribute classes, one that has
     quantity= and unit= (of which <space> and <measure> would both
     be members), and one which has commodity= (of which <space>
     would not be a member).

* Changed the use of &stress; to indicate a stressed syllable to
  <emph> in an example in "Elements Unique to Spoken Texts"

* Changed the example in the tagdoc for <ident> from a GI (kinda
  silly as there already is a <gi> element) to a namespace (not
  necessarily the best example, suggestions welcome).

* Fixed some hard-coded references to be proper <ptr>s

* Fixed a few typos, various other minor things

I plan to check in these changes within the hour. (Can't do so now
due to "disk full" error -- doh!)

Notes
-----
[1] Yes, you might argue that it should be something more limited
    like an SGML NAME so that it could be used in DTDs, too, but no
    one has written or even requested software that produces DTDs
    with prefixes.