When I spoke with Raymond Urgo in January he asked us to consider where the following sorts of information would go:
I assume that Raymond had some sort of classification scheme in mind, such as high, medium, and low. We could create an "importance" attribute for <policy>. The classification tokens for importance should probably be left unspecified.
Revision and review
These are both processes that could be marked up with task.
This sort of information could be marked up with troubleshooting.
Nature of policy
Raymond observed that policy information is inherently abstract since it usually deals with expectations rather than objects. Consequently, any model we come up with must be quite flexible. He pointed out being too prescriptive with our model would limit its usefulness. The work that he has done with clients bears this out. He was emphatic that forms or template-based solutions are too rigid.
In light of Raymond's remarks, I noticed that the section-like elements in the XML that I showed him naturally split into two categories expectations and qualifications on those expectations. This suggests that a wrapper element (e.g., <expectationGroup> might be useful for grouping.
Where to put the <policy> tag?
My current thinking is that <policy> ought to be a section-level element that signifies a policy statement. For example, "An illness lasting more than 5 consecutive work days must be vetted by HR." I've been thinking about this use-case as policy with a small "p." Contrast this with something like a company's hiring policy. This could be a complex collection of policy statements and rationale that would be linked through one or more DITA maps, or perhaps it could be an entire publication. I've been thinking about these use-cases as Policy with a capital "P".
For Policy with a capital "P," tying the <policy> tag to a topic wouldn't suffice in complex cases, although I suppose you could use it as a wrapper topic in a DITA map. However, if you were to use it as a wrapper topic, the semantics would become nebulous. It is far more concrete to tie <policy> to a policy statement like what happens in policy with a small "p."
If we were to tie <policy> to a policy statement, that would also allow us to use a more general name for the containing topic. I've been calling the topic <principle>. Using a topic name like <principle> would also allow us to introduce other expectation-oriented elements at the section-level such as <requirement> or <regulation>.
+1 720 201 8260
Time zone: Mountain (GMT-7)