RE: [dita] Why There are Constraints on Conref

["Ogden, Jeff" <jogden@ptc.com>]

Tony,

I think the description in your message is correct.  In another message
Rob said that you had things reversed, but I think the problem is just
some ambiguity in the wording so that it isn't completely clear what is
the conref source and what is the conref target.

In the html file you attached there is an example of how constraints are
declared using @constraints and @constraints-scope, but I think that
approach is obsolete and has been replaced with a declaration that is
part of @domains.

Your html file also says that you can add and remove attributes, but I'm
pretty sure you can only remove attributes using constraints.

And as Rob says, the reason that you can't just test to see if the
conref material is legal based on the DTD or XSD is that test would be
based upon the content in a particular conref target document instance
at a particular time and at a different time the same conref target
document might be different and might be invalid.  The current conref
validation scheme gives you a guarantee that what is a valid conref will
remain valid into the future.  This guarantee comes at the expense of a
more restrictive policy then is absolutely necessary under some
circumstances.  But the restrictive policy is very similar to the policy
that has existed for a long time with respect to conref and domains and
as far as I know that hasn't been a serious problem.

   -Jeff

> -----Original Message-----
> From: Tony Self [mailto:tself@hyperwrite.com]
> Sent: Monday, September 28, 2009 9:51 PM
> To: 'dita'
> Subject: RE: [dita] Why There are Constraints on Conref
> 
> Hi all
> 
> I have to admit to not previously spending much time trying to
> understand constraints. I am only now trying to get my head around the
> feature, and the recently raised conref issues (which did alarm me). I
> have drawn together a simplistic summary of the feature, prior to
> Michael's proposed "weak"
> addition. I have attached that summary, in case it is useful to anyone
> also playing catch-up. I hope I have got the fundamentals right.
> 
> As I understand it, the main problem is that I can't conref something
> like a simple <note>Don't run with scissors!</note> element from a
base
> (unconstrained) concept topic into a constrained concept topic (even
if
> <note> is allowed in the constrained topic), because conreffing from a
> less restrictive to a more restrictive topic is not permitted.
> 
> Michael's "weak" proposal will impact on the way constraints are
> declared, and the way processors handle constraints.
> 
> Perhaps I'm displaying my ignorance on the issue, but would an
> alternative be for the constraints to be ignored unless the conref
> inclusion "broke" the content model rules of the target topic? So,
> provided the imported <note> would be valid if it were pasted into the
> target, constrained topic, then the conref is valid.
> 
> Hope I'm not muddying the water.
> 
> Tony Self
> 
> 
> 
> 
> 
>