[tei-council] NH revised

[Syd Bauman]

> So, the semantics and expected content of an element can be changed
> just by the attributes it bears?

Yes. As in many cases. E.g. fVal= of <f>:
* <f>a</f>: feature is a
* <f fVal="#b"/>: feature is what is found at #b
* <f fVal="#b">a</f>: feature is unification of a and what's at #b
* <f/>: an error

Seems analogous, although obviously not a 1:1 mapping:
* <q>a</q>: quoted passage is a
* <q sID='x'/>: quoted passage is here to eID=x
* <q eID='x'/>: quoted passage is here (back) to sID=x
* <q sID='x'>a</q>: an error
* <q eID='x'>a</q>: an error


> I find that a leetle hard to swallow, frankly.

I don't understand why. Remember, we're not changing the semantics of
what the element *means* (a quote is still a quotation, an s is still
a sentence), rather the semantics of what the tag means to a
processor: I enclose the entire element vs I indicate the start or
end boundary of the element. 

This is not at all unlike part=[IMF] or next= and prev=.

An <l> alone says "I am a metrical line". An <l> with part="I" says
"I am part of a metrical line, go find the next <l> (which should
have part="M" or part="F") to get the rest of the metrical line".

An <l> alone says "I am a metrical line". An <l> with next="#n" says
"I am part of a metrical line, go find target 'n' (which should be an
<l>) to get the rest of the metrical line".

An <l> alone says "I am a metrical line". An <l> with sID='e' says "I
am the beginning of a metrical line, go find eID='e' (which should be
an empty <l> with no attrs) to find the end of the metrical line".

Not so different, really.


> There is just one tiny precedent though: the <shift> element marks
> the start of something (a segment boundary in fact) when its @new
> attribute is specified, and the end of something when it isn't. But
> its content is always the same.

Yeah, but <shift> is even weirder: it has an application range that I
imagine would be a pain to process properly -- "I am in force only
for utterances of the same speaker, until another <shift> w/ same
feature= but no new= or new='normal'" (did I get that right?)