On 24 Aug 2011, at 15:59, Martin Holmes wrote:

> That still doesn't provide any guidance as to how to handle differences 
> in the way people might choose to use string attributes such as "type", 
> of course. For those, we would have to provide a much stricter set of 
> limited values for e.g. @type when it occurs on <div>, or @type when it 
> occurs on <p>. But even if we did that, I can't imagine how we could 
> convert unconstrained and unpredictable sets of values into our 
> "foundation" values.

my off-the-cuff feeling is that many uses of @type are for analysis,
and can safely be passed on as is via perhaps a @rend. see eg from ECCO:

                        <valItem ident="ActofParliament"/>
                        <valItem ident="Parliamentaryproceedings"/>
                        <valItem ident="abstract"/>
                        <valItem ident="abstracts"/>
                        <valItem ident="account"/>
                        <valItem ident="accountcont."/>
                        <valItem ident="accountofthesociety"/>
                        <valItem ident="accounts"/>
                        <valItem ident="accusation"/>
                        <valItem ident="acknowledgements"/>
                        <valItem ident="acknowledgment"/>
                        <valItem ident="act"/>
                        <valItem ident="actandscene"/>
                        <valItem ident="addenda"/>
                        <valItem ident="addendum"/>
                        <valItem ident="addition"/>
                        <valItem ident="additionalaccounts"/>

These don't effect rendering into "80%TEI", except when we see

   <foo type="x"><bar type="y">blah blah</bar></foo>

and the rule says to turn <foo> into

   <hi type="???"> blah blah</hi>

which is unresolvable.

But we're being quite theoretical here. Would someone like
to show a bit of TEI which falls into the 20%, ie outside the
semi-mythical "core", and which one would have to dumb down
to make it "80%" compatible? 

I may note, by the way, that processing ODD documents to
make documented forms (HTML and PDF) go through this
process, turning weirdo things like <specList> and
<elementSpec> into core TEI, before trying to format them
(odd2lite.xsl, if you care). Works fine. 

NZETC do something similar with their texts, I understand.
Sebastian Rahtz      
Head of Information and Support Group
Oxford University Computing Services
13 Banbury Road, Oxford OX2 6NN. Phone +44 1865 283431

Sólo le pido a Dios
que el futuro no me sea indiferente