[tei-council] att.sourced (<lb ed="1674">) contradiction

Syd Bauman Syd_Bauman at Brown.edu
Sat Jan 5 19:21:41 EST 2013


The difference between data.code and data.pointer is that in some
previous (probably pre-release) iteration of P5, data.code was
intended to point to something *in the same file*, thus being a
mechanism for on-the-fly controlled vocabularies.

The TEI has (or at least had) 5 general-purpose mechanisms for
providing vocabularies for attribute values:

* list of values is supplied in the ODD (either TEI ODD, user's ODD,
  or combination thereof) and thus the schema: data.enumerated.

* list of values is supplied in the instance file (typically in the
  header): data.code.

* list of values is supplied in a specific external source:
  data.language (which I think is only example of this at the moment,
  since data.outputMeasurement and data.sex do not reference the
  specs, but rather check for values themselves)

* list of values is supplied somewhere on the web: data.pointer.

* list of values is magically generated at run-time by the user, with
  the algorithm supplied in a TEI-conformant way (@cRef) or some
  other way, typically an index into an external database (@key).

I think Lou is right, most people who actually use @ed are perfectly
happy using a data.word-like string that everyone in their project
(sometimes in their entire field) recognizes immediately. IMHO,
that's an argument for this to be data.enumerated.

But I can see that a lot of folks would want to point from @ed to a
bibliographic entry for the particular edition. In which case
data.pointer seems reasonable. So I'm thinking (reiterating some of
this thread) we have three choices:

* change @ed to data.pointer, thus requiring (in some loose sense)
  that the "Silver" edition be indicated with "#Silver" and that the
  1674 edition be indicated with "#e1674" or some such. (But as has
  been pointed out, many users obliviously point to non-existent
  local files named "Silver" and "1674", and go on with their lives
  quite happily.)

* change @ed to data.enumerated, and folks who want a bibliographic
  description of their edition can put it inside the <desc> of the
  appropriate <valItem>.

* change @ed to data.enumerated, and provide @edRef as a
  data.pointer.

I don't really see any place for data.text here.


More information about the tei-council mailing list