[tei-council] persName question

[Lou Burnard]

Sebastian Rahtz wrote:
> Whatever else you do don't abuse semantics....
>
> The "mark it simply and canonicalize it later" is OK, but
> you'd still want to use a new namespaced attribute,
> I reckon, for sanity.
>
> What about two other cases:
>
> a)
>    <persName>John Smith</persName>
>
> where we are not sure _which_  John it is (we
> know of three). In this case <persName ref="#smith1 #smith2">
> would have a different meaning (ie alternation rather than
> a list)
>
>   

This is a completely different problem. The point of using @ref and @key 
is that you *have* resolved the target (or claim to). If you want to 
mark up uncertainty about the resolution then you use the wonders of the 
certainty module. Can you cite *any* other case where we use  multiple 
values for an attribute to indicate an alternation rather than a 
concatenation of some kind? I certaintly cannot.

> b)
> <persName ref="wikipedia.org#smith www.bbc.co.uk/people/smith">
> where there are two useful and distinct pointers to the same person.
>
>   

Again, this seems to me an extension beyond what the current attributes 
are designed to support. Yes it might be nice to use @ref to point to a 
variorium of further info, but that's not what it's for. If you want to 
do this, put the other references inside your <person> element.

> So does a multiple @ref mean
>
>   #1, #2
>   #1 | #2
>   #1 & #2
> ?
>
>   
it always and only means the first of these


> how would  we distinguish?
>
> You _could_ organize your family into
>
>    <listPerson xml:id="Simpsons">
>       <listPerson xml:id="ParentSimpsons">
>           <person xml:id="Home">...</person>
>           <person xml:id="Marge">...</person>
>        </listPerson>
>       </listPerson xml:id="BabySimpsons">    
>          <person xml:id="Maggie">...</person>
>       </listPerson>
>   </listPerson>
>
> what would be the problem with that?
> the fact that Marge would also occur
> in the Bouvier family?
>
>   
Yes, exactly.