Question about forward key references (issue 12007)

The ad hoc group that is working on the Indriect
reference/keyref proposal (issue 12007) needs some feedback from the larger
DITA TC.

 

The issue is how important is it to be able to reference a
key before it is defined? Or stated another way, how important is it to allow
forward references?

 

Consider the following simple example where a topic that
uses a key reference is referenced from two different maps:

 

<topic  id=”topic1”>

   . . . 

   <xref
 keyref=”mykey”/>

   . . .

</topic>

 

<map id=”one”>

   <keydef
 keys=”mykey”  href="”www.google.com”/>

  <topicref 
href="”topic1.dita”/>

</map>

 

<map id=”two”>

  
<topicref  href="”topic1.dita”/>

   <keydef
 keys=”mykey”  href="”www.google.com”/>

</map>

 

Here <keydef> is a specialization of topicref that
just defines a key name and an associated reference. It allows a map author to
supply or override a reference in a topic (or in a map) when the topic (or map)
uses key references.

 

The question we need feedback on is if it is important that
the map two case work and produce the same result as the map one case, or is it
OK if in the map two case that the key reference in topic1 is undefined and is
treated as an error.  The question probably isn’t very interesting
for such small and trivial examples, but for larger and more complex examples
there are pros and cons to allowing or not allowing both cases.  Our ad
hoc group would very much like to learn of use cases that require or would
benefit from allowing forward references.

 

Please send use cases and other comments to the DITA TC list
or to me () before the end of
the day this Thursday, 6 December.

 

If anyone wants to look at a draft of the entire proposal
for issue 12007, it is posted on the OASIS DITA TC site.  See: http://www.oasis-open.org/apps/org/workgroup/dita/download.php/26282/IssueNumber1207v6.html

 

Thanks!

 

   -Jeff