[tei-council] death of <*Struct>: pending problems & suggested solutions

[Syd Bauman]

Per my action item from 07-25 ...

For each issue I've listed a problem or question, each (save the
last) along with an answer or solution. Unless there are objections
to an answer or solution, that's what we'll do. The only issue with
an answer that I think merits serious discussion is Ranges.

Zone=
-----
Question: Should the zone= attribute of <timeStruct> be retained as a
          new zone= attribute of <time>?
Answer: No.

Type=
-----
Question: Should the type= attribute of <dateStruct> and <timeStruct>
          be retained?
Solution: Yes; make <date> and <time> members of class att.typed (I
          am the only one who complained that subtype= was overkill, so ...)

Ranges
------
Problem: We need to think through <dateRange> and <timeRange> with
         respect to data.duration, ISO 8601 range formats, and
         notBefore= & notAfter=.

Proposal: * delete <dateRange> and <timeRange>
          * add <date> and <time> to att.datable (thus they get
            notBefore= and notAfter=)
          * alter att.datePart so that rather than having
            attribute value { data.temporal | data.duration }
            it defines two attributes
            attribute value { data.temporal }
            attribute dur { data.duration }
          * add Schematron rule to <date> and <time> that insists
            that either ( value= and/or dur= ) OR ( notBefore= and/or
            notAfter= ) is present, but not both

          ALTERNATIVE: 
          Rather than having a Schematron rule, could plop all four
          attributes into alternate pairs in a single class. At least
          I think this is supposed to work, but at the moment it
          generates an invalid schema. (I've sent Sebastian separate
          mail.) 

          Either way, this potentially has significant repercussions,
          and is worthy of discussion.

<offset>
--------
Question: Should <offset> be a member of att.datePart?
Answer:   No, remove it.

Full= of att.datePart
----- -- ------------
Question: Should full= exist?
Answer:  No, kill it.

Type= of att.datePart
----- -- ------------
Question: Should <day>, <distance>, <hour>, <minute>, <month>,
          <occasion>, <second>, <week>, and <year>, have a type=
          attribute?
Answer:   No, remove type= from att.datePart.
          (Note: if they *should* have type=, they should get it from
          att.typed, although I still am queasy about subtype=.)

Precision, accuracy, exactness
---------- --------- ---------
Problem: We still need to sort out this precision/exactness stuff.