[tei-council] the "key" attribute

[Syd Bauman]

The problem (ti seems to me) was caused when we overloaded the
semantics of the attribute name "key=" when son of ODD was born. I'm
inclined to say the best way to fix it is to fix the original error,
and rename the attribute used by <memberOf> et al. 

That is, there are three kinds of semantics:

* I am a pointer, and point to a digital resource via a URI

* I am a pointer, and point to an element inside this current
  document by matching the value of its (possibly non-unique) ident=
  attribute

* I am a key into a database

In cases where there are no special semantics (like next= and prev=)
we use target= for the first, and key= for both the second and third.
I think we need separate names for the 2nd and 3rd. My conservative
instinct was to leave the P4 use for names as key= and change the
name used by ODD elements. But in fact, since we are permitted to
"break" existing documents, and it is just another template in the
transformation from P4 to P5, there really isn't any strong reason not
to use key= for the ODD elements and make up a new name for the
database looker-upper attribute for names.


While I'd be fine with adding a new attribute in the naming contexts
that is explicitly a URI reference to the regularization or
disambiguation information, I would be very hesitant to completely
remove key= and thus the capability to refer to databases that are
not addressable via URI. (As I said above, I have no problem renaming
it; nor do I currently have any opinion on whether if there are two
attributes, one for database record identification and one for a URI,
they should be mutually exclusive or not.)