[tei-council] comments on edw90

[Syd Bauman]

My apologies for the delay. Several pieces of e-mail seem to have
gotten swallowed in my Inbox. I'm only going to get to this one
tonight, I'm afraid.


SB>The idea from Paris was that a <valList>, whether closed, open, or
SB>semi, is required for an attribute of type tei.data.token. I still
SB>think that's fine.

LB> No, I think the presence of a valList implies that the datatype
LB> should be tei.data.enumerated, which maps onto either an explicit
LB> list of alternatives (if the valList is closed) or onto
LB> tei.data.token otherwise.

Sorry -- You are right, I had meant "tei.data.enumerated" in that
sentence.


LB> [That datatypes should all be mapped to xsd: datatypes] is simply
LB> reiterating a decision we took at the Council meeting in Oxford,
LB> if I remember aright.

The minutes say only "it was proposed [within META] ... to define
datatypes for attribute values more abstractly, mapping them to W3C
schema datatypes where possible".


LB> So the choice is really
LB> 
LB> (1) use xsd:duration (which includes units) , and thus remove all
LB> the additional attributes that specify units
LB> 
LB> (2) use plain old numeric dataype of some sort, and retain the
LB> unit attribute
LB> 
LB> Choice (1) is only going to work for durations where we can be
LB> confident that the xsd units are going to meet TEI needs,
LB> obviously. I fear there may be fewer of these than anticipated --
LB> probably people will be content to use SI units for times, but
LB> you can bet someone will always want to measure (e.g.) distance
LB> in rods poles and perches.
LB> 
LB> At present the only attributes assigned tei.data.duration are
LB> age= and dur= Choice 1 fits fine with the latter, and reasonably
LB> well with the former, imho. Any others?

SR> surely the point of these attributes is exactly to standardize
SR> described text? so when I see "67 rods", I want to, and should,
SR> encode it as <foo distance="32.3m">67 rods</foo>?

SR> choice (1) seems obvious to me, and always has done. I don't
SR> think SI units do the job, I'll lobby ISO....

I'm not entirely sure I fully understand what Lou's getting at, but I
think I agree with Sebastian. Are we talking just about durations,
here, or about any and all things that have units?

I haven't done an exhaustive search, but the attributes that are
concerned with duration include
* dur= of %tei.timed; & <recording>
* age= of <person>
* interval= & unit= of <timeline> & <when>
and could include 
* value= of the various date & time elements

While I think sticking with a number-and-unit-in-one approach of
xsd:duration probably makes sense for all of these, there are some
problems with using just xsd:duration alone for them.

First, xsd:duration permits negative values, which don't make sense
for age= or interval=/unit=. Second, it is my bet that TEI users in
general are not going to want to express someone's age as, e.g.
"P36Y", but would much rather "36 yr" (and we can argue over details
like "year" vs "yr" vs "years" vs "yrs", white-space or not).

IMHO we are *not* talking about <measure>, where the unit should
almost assuredly be kept separate from the numeric component.