Re: [dita] Why There are Constraints on Conref

["Rob Hanna" <rob@ascan.ca>]

The problem is the other way around. A constrained element cannot conref an unconstrained element as the substructure in the unconstrained element may break the rules defined for the constraint. 

For example, an unconstrained paragraph will allow for a list element whereas a constraint may disallow a list within a paragraph. By conreffing the unconstrained paragraph into the constrained paragraph element, the author may be introducing an invalid substructure. Even if the paragraph is valid according to the rules defined within the constraint at the time the conref is included, the unconstrained paragraph could be changed at anytime to include a list element thereby invalidating the content within the constrained paragraph element.

By introducing the mechanism for a "weak" constraint processors will be permitted to ignore the constraint with a warning if the conreffed content is unconstrained. This means that an author would be unable to break the constraint while creating the content but could conref content that was written against another less constrained model.

If the constraint is created as "strict", the author would not be able to conref the less constrained content.

By introducing the constrained task model in both ditabase and the task DTDs, authors will be able to conref from task elements created using the unconstrained task model.

This in effect makes the constraint more of an authoring guideline than a rule. It will promote consistency while allowing for exceptions.

I hope this is clearer and that I have properly interpreted the facts thus far.

Cheers,
Rob Hanna
Sent from my BlackBerry device on the Rogers Wireless Network