RE: [dita] Why There are Constraints on Conref

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

Constraints potentially offer a unique opportunity for managing different
business rules across an enterprise or industry. In my opinion, a task is a
task - the current proposal seems to change that. Again in my opinion, this
is fundamental to exchange of information across groups. What we are
otherwise talking about goes well beyond the simple task example provided in
DITA 1.2. Architects could very well try to take advantage of the constraint
mechanism as a means of improving the usability of the authoring templates.
In so doing, under the current proposal, they would be rendering their
content unusable outside of their environment.

If one group imposes business rules that are stricter than another, that
group must decide how it wants to reuse content from the other group.
Ideally, the two groups can reach a mutual understanding and twin their DTDs
to match. In reality, the business rules may be different for reasons beyond
the control of one group or another.

The worst possible result would be that the first group must rewrite and
manage the content provided by the second group because it was incompatible.
There are inevitably situations where this will occur - but we have an
opportunity to minimize the frequency of these situations.

The opportunities for reuse between constrained and unconstrained content is
significant. Unlike examples cited where reference content is reused in task
content, the two variations of constrained and unconstrained information are
semantically similar. The failure or fall back mode of processors to handle
unconstrained content reuse within constrained topics should not be the same
as reusing dissimilar content types.

I believe that authors and publishers should be alerted to instances where
the reused content fails to adhere to the stricter version of the model but
they should not be prevented from using it.

What if we introduced a sort of name space mechanism to the conref that
could be used by processors to validate the content against the proper
model? This would only work between two topics of the same document type
(one loose and the other strict).

Cheers,

Rob Hanna