This is a successor to ticket http://purl.org/tei/bugs/3413346. Following the decision of Council at the FTF in Oxford in September 2012, the original ticket is to be closed and a clean new ticket opened to discuss the deprecation of data.key.
The option of using and documenting private URI schemes is now implemented in the Guidelines and will be in the first release in 2013, so we are one step closer to being able to plan a deprecation strategy for magic-token attributes. data.key itself is now gone, and attributes which used to have that datatype (such as @lemma, @key) have now been redefined as having data.text (rev 10512, http://purl.org/tei/bugs/3531765\). So:
- Should we deprecate the use of @key (as defined in att.canonical, distinct from its definitions in memberOf etc.)?
- What exactly should we encourage instead, in the various contexts in which @key is used and exemplified? (There are several dozen examples which use att.canonical's @key, all of which would need to be replaced.)
- Should @lemma also be up for consideration? (Probably not. See the discussion on http://purl.org/tei/bugs/3413346.\)