[sc34wg3] XTM 1.1 issues

[Lars Marius Garshol]

* Martin Bryan
>
> XTMDM would show its origins

That sounds rather bizarre to me. TMDM was constructed to be the  
model that was missing from XTM 1.0, and has taken a lot of flak over  
the years for being too faithful to XTM 1.0. So now that we  
synchronize XTM with the result of the model work over four and a  
half years, you want us to name it after the model that was created  
for XTM? If we wanted to change the name to show the origins, why not  
XTMXTM?

The changes that have been made in the new XTM version are there for  
two reasons:

   (1) We've had five years of experience with XTM, and the result of  
that
       is that some of the things that seemed like a good idea in  
late 2000
       no longer do. (Like XLink support, which I argued strongly for  
back
       then.)

   (2) Developing the model has given us new insights into the  
concepts the
       syntax is meant to be a representation of, simply because those
       concepts were not fully developed at the time the syntax was  
created.
       Now that the concepts have been fully developed, we see that some
       things in the syntax can be improved on, and so we're  
improving it
       now. (This is why people were arguing so forcefully back in  
2000 that
       the model should have been developed first. Had that happened  
the last
       five years would probably have looked rather different.)

To me, this really is just another small step onwards from XTM 1.0. I  
can't see any reason why we should change the name. If we'd come up  
with something like TM/XML and proposed that as XTM 1.1 (or 2.0) I  
would have understood the objections, but not with the current draft.

> and show it was an extension to the original concept.

Well, the extensions are types on names and datatypes. That's not  
much of an extension, is it?

> Coming at things from the other direction, I'm struggling with  
> subsetting
> ontologies that are ridiculously large (millions of RDF triples).  
> If the
> ontology had been properly designed in a modular format things  
> would have
> been infinitely easier. Long live modularization features.

Well, <mergeMap> is staying, so they're not dying yet.

(BTW: I've been able to convert that ontology to a topic map, but the  
untimely demise of my previous laptop has prevented me from loading  
it into the RDBMS backend so it could be visualized with a custom  
Navigator application. Still hoping to be able to do that.)

--
Lars Marius Garshol, Ontopian               http://www.ontopia.net
+47 98 21 55 50                             http://www.garshol.priv.no