RE: [dita] Normative language specification

["Grosso, Paul" <pgrosso@ptc.com>]

> -----Original Message-----
> From: Eliot Kimber [mailto:ekimber@reallysi.com]
> Sent: Wednesday, 2010 January 13 16:45
> To: Bruce Nevin (bnevin); Ogden, Jeff; dita
> Subject: Re: [dita] Normative language specification


> 
> Again, if there is a out and out conflict between what the prose says
> and
> what the DTDs say, then there is a bug in one or the other, but until
> the
> bug is fixed, the prosed is presumed to be correct. For example, if
the
> prose says "you can have zero or more foos" and the DTD requires at
> least
> one foo, until the prose is corrected (assuming it's wrong in this
> example),
> a document with zero foos is, by definition, a conforming document for
> this
> particular rule.

Right, but is an application that uses the DTD and gives a validity
error for the document with zero foos a conforming DITA application?

Per your words, it is not.

So what should the implementor who is producing an out-of-the-box
DITA application supposed to include in its distribution?  The
broken DTD that will complain about the valid document or a
privately corrected version of the DTD that isn't what OASIS
is distributing?

I'm not that worried about the issue, but that, I think, is Bruce's
point.

paul