[sc34wg3] CXTM Issue: Values

[Lars Marius Garshol]

* Lars Heuer
>
> If my Topic Maps engine reads XTM / CTM / whatever and the source
> contains "0"^^xsd:boolean for an occurrence value and the engine
> returns "false" for the value of that occurrence if the user requests
> that value, this would be wrong according to the given Topic Maps
> standards, since the engine does not keep the "original" (string)
> value?

No. :-)

TMDM doesn't actually say what your API is supposed to return. It  
can't do that, of course, and this is why it has no conformance  
section. TMDM on its own doesn't do anything. It's only when you  
couple together, say, CTM -> TMDM -> CXTM that you get some concrete  
ruling on issues like this one. And in this case, with the 2008-04-14  
CXTM draft, outputting "false" would be wrong.

Returning boolean false in your API, on the other hand, would be the  
best possible way to represent this.

> Well, the CXTM tests are not part of the standard, right? My feeling
> is, that the spec. should be explicit and self-contained; so (3) seems
> to be the cleanest solution here.

Yes, I agree.

I propose then that:

  (1) CXTM should explicitly recognize and canonicalize values belonging
      to the types recognized by CTM and TMQL.

  (2) The test suite will avoid testing canonicalization of values of
      other types, to avoid penalizing tools which add specific support
      for these.

Unless I hear anything to the contrary I'll do just this in today's  
draft.

--Lars M.