[tei-council] persName question

[James Cummings]

'persName' picks up a @key, @ref, and @nymRef from att.naming.

http://tei.oucs.ox.ac.uk/P5/Guidelines-web/en/html/ref-att.naming.html

@key is defined as data.key = any unicode string
@ref is defined as a single data.pointer = any URI
@nymRef is defined as 1–∞ data.pointer = any number of URIs.

I'm in a situation where a project wants to point from something like:

<persName>Mr and Mrs Smith</persName>
or
<persName>The Smiths</persName>
or
<persName>The Smiths and little johnny</persName>
or
<persName>The Smiths and little johnny and his 3 sisters</persName>

well, though maybe those last two are stretching it.

What I'm wondering is why @ref and maybe @key are also one to infinity like 
nymRef?

What they want to do is on one 'persName' point to more than one individual 
'person' element.  I.e. they know stuff about both Mr and Mrs Smith and 
want to record their information separately.

Now, They could point instead to a nested 'listPerson' which then contains 
both 'person' elements, but how do you differentiate that in this instance 
it is Mr and Mrs Smith, but in this other instance just Mrs Smith and kids. 
  They end up with lots like the second example 'The Smiths' meaning 
usually just the adult members (which change over time). And there are so 
many other possible nestings of listPerson, how do you cope for them all?

I'm just trying to remember what the logic was in allowing nymRef to have 
more than one URI but having ref only allowed to have one?  I am sure I 
could think up many cases where a single name referent refers to multiple 
people (who are not treated as a group, so personGrp is not the answer).

In cases such as this I don't think pointing to a single listPerson which 
groups multiple person elements is the right answer (because then you have 
to repeat multiple listPerson elements for different groupings).

Wouldn't it be a lot better to allow:

<persName ref="#SMI01 #SMI02">The Smiths</persName>

Other options open to me as I see it:
a) abuse the whitespace allowed in @key (oops maybe not)
b) add an entire new attribute (good but I wanted to avoid extra namespaces)
c) use some abusive local encoding format that I then canonicalise to a 
pure TEI solution at a later date.


Thoughts?  Am I missing something obvious?

-James

-- 
Dr James Cummings, Oxford Text Archive, University of Oxford
James dot Cummings at oucs dot ox dot ac dot uk