[tei-council] Conformance .... the continuing saga

[Lou Burnard]

Syd Bauman wrote:
> While on the face of it Lou's suggestion (mention all 3 methods of
> generating a complete TEI schema: via ODD, parametrized DTD, or
> stitching RNG) seems quite reasonable, Arianna's point that we would
> then need to make sure SG explained both DTDs and RNG enough to do
> this is a strong argument against, IMHO.

Depends what "it" is. It only has to explain enough for the further 
explanation in MD to be comprehensible.

> 
> SG should not mention details of DTDs at all. (And yes, Lou,
> stripping 'em out is still on my plate). Nor should the rest of the
> Guidelines rely on DTD syntax at all.


When you get around to reviewing the mail on this list following 
Arianna's work on that chapter, you may (possibly) notice that this 
issue has already been raised there.


> 
> There are lots of non-conformant methods of making a schema that will
> validate TEI documents that could be conformant.[1] I do not think
> the Guidelines themselves should discuss any of them. The Guidelines
> are already daunting enough. Why add large sections of potentially
> confusing information that will likely be used by an extraordinarily
> small number of users (perhaps 0), to explain something that isn't
> even conformant?

It is your assertion, and your assertion only, that no one will be 
interested in making a schema capable of validating TEI conformant 
documents without using ODD. You may be right, you may be wrong. But 
simply repeating the assertion doesn't make it true. Nor does it address 
the counter arguments raised in my earlier posting on this topic.

> 
> Again, a separate paper or HOWTO explaining this may be a good idea.
> 
> 
> BTW, does anyone actually know how to go about stitching together the
> RNG schema fragments in a manner that permits similar capabilities to
> using parametrized DTDs?

Anyone capable of reading the output from Roma maybe?

  We would have to figure out how to do it
> (although, because this is not conformant, I daresay we do not have
> to figure out how to do it optimally), do extensive testing, and
> write up a serious chunk of prose instructions. 

Yes, that's what I said earlier.