[tei-council] how much can one customize and still have a TEI document?

[James Cummings]

Lou Burnard wrote:

> > The TEI editors' best attempt at this is currently the chapter on
> > conformance. This states, as already noted in this thread, that you are
> > not allowed to remove the TEI Header, and furthermore specifies three
> > topics (identification, publication, source) that your header must
> > address in some way.

While I don't in any way disagree with this, I just want to check if these are
all still things we want to insist upon as part of our levels of conformance?
What if someone wanted to store the header as a whole as an external resource
and just point to it but otherwise be a normal TEI document?  Do we still want
to specify those three topics above as required?  Obviously I'm just trying to
think of use cases where someone would want to not have these but still really
the document could be considered TEI.  I'm not really being successful, but
playing devil's advocate.  I'm definitely interested in exploring the use of TEI
document fragments which when assembled form a proper TEI document, but
separately (since missing a header) do not.

> > The "nonformal wording" David refers to was introduced (probably by me)
> > to alert people to the technical possibility of using the extension
> > mechanisms to do some dangerous/illegal/ill-advised things, not to
> > licence such madness!

I in no way want to duplicate docbook via an ODD.  It just occurred to me as an
extreme which was much more conformant than someone accidentally deleting the
@xmlns from the TEI element or similar.

> > Questions I think we need answers to:
> >
> > (a) what mechanisms beyond a more formal statement of the limits of
> > modifiability are feasible?

Are there any which are reasonably feasible?  I have no objections to the true
test of conformance being a formal statement in prose in the guidelines.

> > (b) and what, beyond what's already on the table, should those limits be?

That's what I'm not sure about.

The two principle things I think a TEI document should have or do are:
a) TEI namespace
b) validate against a schema that is generated from a valid TEI ODD.

after that, I'm less picky about whether it has a teiHeader.  Obviously I think
*some* metadata would be an advisable thing, but it makes it just that more
complicated for fragmentary documents to be considered TEI documents.  But this
is really a non-existent problem isn't it?  I mean I use my fragmentary
documents perfectly fine with my customised ODD, then validate the full document
against tei_all and I'm fine and perfectly conformant.  I'll stop wittering
nonsensically then.

-James

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