On 10/11/16 23:29, Syd Bauman wrote:
* First thought on solving the problem Hugh posted is to just drop support for DTDs and XSDs. :-)
Not really an option
* I'm yet to be convinced that <rngContent> is a good idea, let alone a requirement to move forward, but have not thought about it very much yet -- am willing to be convinced.
Here are my reasons for thinking we should do it. There are plenty of existing ODDs around which use RELAXNG to express content models. We'd like them to move over to using pure ODD, but we can't expect them to do so overnight. Their maintainers/users won't however ever start thinking about it until something changes to bring that possibility to their attention. Hence the suggestion of (effectively) transforming <content> to mean <pure-TEI-content-only>, and providing <rngContent> as a simple transition mechanism. We can discuss later whether we want to move to a deprecation of <rngContent> in a year or two (or an addition of <xsdContent>) when we see how things playout. In the meantime we are (a) making things technically a lot simpler (b) making things a lot more transparent.
* Why would we want to drop support for RNG in content models? Lou wasn't suggesting that initially, and my instinct is that's a bad idea. (Adding XSD is not an idea I'm fond of, but it's better than dropping RNG.)
I wasn't suggesting dropping support for RNG in content models. I am proposing making it explicit rather than letting it sneak in by the back door.
* Remember, the solution to the dreaded dreaded double-declaration- of-id problem is to turn DTD compatibility mode OFF. (We don't want the limitations of DTDs imposed on our RELAX NG.)
Still not really an option, in my view.