On Tue, 2013-04-09 at 15:39 -0400, Simon St.Laurent wrote:
> On 4/9/13 3:21 PM, John Cowan wrote:
> > When you publish a set of documents, publishing the schema to which you
> > claim they conform is very useful for everyone.  For one thing, it's
> > concrete, testable documentation about what to expect from you.  This is
> > the opposite of the "normal" use of schemas as input validation: here
> > they are serving as *output* validation as well as documentation.

>
> That qualifies as a neat reversal of polarity.

It's been fairly common in some circles for decades (starting with
SGML). I gave a paper in this area at SGML 96.

One reason is that markup is often used for transcriptions of
already-extant documents, whether cuneiform on fragments of pottery or
printed books from the 1700s or 1800s, or brands on people's rumps :-)

If the DTD or Schema conflicts with what's in the primary text, the
primary text is what's right, not the schema. This is referred to in the
literature as descriptive markup.