[tei-council] header element to reference schema

[James Cummings]

Sebastian Rahtz wrote:
> We have an outstanding feature
> request to extend the TEI header  to allow a pointer in the TEI
> to the schema (in the loosest sense) against which this document is valid.
> (http://sourceforge.net/tracker/index.php?func=detail&aid=1650195&group_id=106328&atid=644065)
> 
> 
> The requestor boldly begs " this should be able to cope with different
> namespaces... i.e. I should be
> able to reference the relaxNG schema for elements starting with rng: the
> tei: schema I'm using, the svg: schema I'm using, etc.  I should be able
> just to point to an ODD, an RNG, etc. or embed my ODD
> etc."

I didn't think I was being that bold.  I just felt that if we implemented a
solution one should be able to embed the schema(s) or point to them.  I.e. point
to my ODD, embed its RNG, and point to some other publicly maintained schema
(say the SVG schema).  Basically I'm just looking for a point at which to do
this in the header, probably under encodingDesc.

> My inclination is to say that ISO NVDL is supposed to cope with this,

I don't know NVDL well enough, but a quick glance at it seems like it might do
what we want.


> Regarding the addition of a header element (or PI or whatever) to the TEI
> scheme to point to a scheme, choose one of the following:
> 
> B. I think it is really important and the TEI must deal with it  by
> itself for 1.0
> D. It can be dealt with using NVDL and we should document that somewhere

I don't see these two as mutually exclusive.  I think it is important, and TEI
should discuss it and document at least one recommendation on how to do it.  I
don't view 'deal with it by itself' as excluding using existing standards.  It
may be that NVDL should be the recommendation we choose.  The only things I
could find about it that bothered me were that in all the examples in the
'annexes' of the ISO publication, it uses Relax NG Annotations for its
documentation...so that means declaring yet another namespace...

I'm assuming that NVDL has no problem if it points to a schema language or
similar that whatever is processing it doesn't understand. I.e. if I point to my
ODD, then oxygen won't die, just say it doesn't know about that kind of thing.

Similarly, this only partly answers the original spec.  This gives us a way to
point to external schemas, but no recommendation on a manner to embed the same
in your instance document.  (i.e. Say I want to stick my ODDs, automatically, in
an instance document so that the two are inseparable.)

-James

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