Potential Issue: When Does Applicability Apply to Key SpaceDetermination?

This is an order of processing issue, although we try to avoid mandating processing order or methods in favor of stating what the required or desired result should or must be.

  We didn't address order of processing issues or the effect of conditional processing on Key References when we voted on issue #12007.
We probably should have, but we didn't.

  The DITA spec.
has never really talked about order of processing issues or the required results.

  Order of processing issues by their nature are interrelated.
I don't think that you can mandate one aspect of order of processing issues without considering the others (how do cascading attribute values from map to map effect conditional processing and how does conditional processing effect key references to pick one example or how does the timing of conref processing effect cascading which effects conditional processing which effects key references to pick another).

  The DITA 1.0, 1.1, and so far 1.2 specs.
have not addressed these issues.
Adding them to DITA 1.2 really requires some thoughtful consideration, a proposal, and ultimately a vote by the DITA TC.
That will take time and effort.
If we want to take that time and effort now, then we can consider making this a MUST item.
If we don't, the spec.
can either remain silent on the issues or we can make it a SHOULD item.
SHOULD is my preference.
It gives everyone a statement of direction without putting implementations that were in compliance out of compliance.

  You can look at this in at least two different ways:

  1. Making this a SHOULD rather than a MUST requirement is a failure; or

  2.
Making this a SHOULD requirement rather than a MUST is an improvement that can be further improved in DITA 1.3.

  By the way, I am not making this argument because of anything that the Arbortext Editor implementation does or does not do as far as Key Reference processing.
I believe that we will implement conditional processing and Key References to meet the SHOULD requirement.
I am less sure that our key reference processing will be correct with respect to all other order of processing issues (cascading, conref, conditional processing, …), simply because the spec.
is silent on those issues and so I’m not sure that the desired behavior is.
We have made what are hopefully informed guesses and implemented something.
I can imagine that once the TC states what the desired behavior for all of this is, that we’ll need to make some changes.
It is unclear how much work will be involvedand therefore how quickly any changes can be made.
Or perhaps we’ll be lucky and our guesses will turn out to have been right and we won’t need to make any adjustments at all.

  -Jeff

  >