[sc34wg3] XTM 1.0-style reification and TMDM

[Lars Marius Garshol]

Lars Heuer brought up the issue of what happens when in a topic map  
some construct and some topic have the same URI as item identifier  
and subject identifier, respectively.

In XTM 1.0, this means that the topic reifies that construct. The  
TMDM does not mention this issue at all, and neither does XTM 2.0,  
since this is not an issue in XTM 2.0. However, we need to have an  
answer on the data model level, where it applies irrespective of what  
syntax(es) the various parts of the data originate from.

Having thought this over, my opinion is that in TMDM this doesn't  
mean anything, just as it doesn't mean anything in XTM 2.0.

However, the question is what we do with XTM 1.0. We could say that:

  (a) Reification is evaluated when we see a <subjectIndicatorRef/> that
      points to the item identifier of some non-topic item. If this  
is the
      case we have reification. The trouble is that this will give  
different
      results depending on the order of processing, so this doesn't  
work.

  (b) XTM 1.0 is not our problem. All XTM drafts up until the one  
following
      the Atlanta meeting covered the new XTM as well as 1.0, but  
after that
      we abandoned that policy, and now XTM 1.0 is only defined in the
      old TopicMaps.org spec. (And what that means is anybody's guess.)

  (c) Reification is evaluated at the end of the XTM 1.0 import. That  
is,
      if there are any matching SI/II pairs, this is considered to be
      reification. This is probably expensive to implement, however, and
      a bigger issue is that SI/II pairs that didn't originate in the
      XTM 1.0 file will also turn into reification.

  (d) We could extend (a) and say that reification is *also* evaluated
      whenever -id- attributes occur on non-topic elements. This would
      rid us of the ordering problem, and should work for all imaginable
      combinations.

I think I have convinced myself that we can stick to the line that  
this doesn't mean anything in TMDM, and that the old XTM 1.0  
reification can be confined to XTM 1.0. The only problem with this is  
that we have no official document to write this up in, but then we  
have no official document specifying the rest of the XTM 1.0  
interpretation, so I guess this isn't really an issue.

--Lars M.