[tei-council] namespaces and modifications

[Syd Bauman]

> Hrmmm. If I've added a brand new element, my:foo, I've done so in
> my own namespace, of course. But if I add it to class att.typed, it
> joins that club and gets the TEI element tei:typed, right?

This turns out not to be the case. Being added to class att.typed
gets you the TEI attribute type=, *not* the attribute tei:type=.
Attributes, unless explicitly qualified, are in no namespace.

"The namespace name for an unprefixed attribute name always has no
value." -- http://www.w3.org/TR/REC-xml-names/#defaulting


But Lou has pointed out some of the major headaches I was referring
to when this came up. Things like the following.

* What if user adds new attribute to existing TEI element? Does she
  have to put it in her namespace, or can it be in no namespace like
  the other attributes?

* What if a user deletes attributes from a standard TEI element --
  does that element now have to get moved from the TEI namespace to
  her namespace? Especially if it was a required attribute, normal
  TEI software may not be able to process that element anymore.

* If she deletes an attribute from an attribute class, do all members
  of that class now have to change namespace?

* What if a user adds an existing TEI element to an existing TEI
  attribute class, such that it now has attributes it doesn't have in
  vanilla? 

* What about the reverse: she removes an element from an existing
  attribute class?

And we haven't even started to think about changing model classes and
content models yet!

Issues like this are why I think this general idea that putting new
elements in a different namespace will help in processing TEI
documents is tenuous at best.