[sc34wg3] Problem with wildcards

[Lars Heuer]

Hi Lars,

[...]
> If we ever decide to merge sweden.ctm and norway.ctm Oslo and
> Stockholm wind up merging, because they both have the same item
> identifier:
>   file://.../geo.ctm#$__1.capital

Good catch. That's a problem, especially if we have some normative CTM
files like the TMCL ontology which are utilized by many topic maps.

> One possible fix might be to change the rules for %include so that
> only the item identifiers from the including master file are
> preserved, and the ones from the included file are lost.

Well, we have had a similar directive which would do nearly the same
thing: The import-directive
<http://www.itscj.ipsj.or.jp/sc34/open/0880rev.htm#dir-template-import>.
That directive imports only the templates and the wildcards would be
resolved against the master file's document IRI.

I was never a big fan of the include-directive and favoured always the
import-directive, so I'd support the proposal to drop the item
identifiers of the included file.

> Or maybe we have to give up the goal of predictable item identifiers
> for wildcards, although that would suck royally, as a lot of the tests
> in the suite rely on this.

As said I support the removal of the item identifiers of the included file.

Anyway, this was my first reaction on this, but I believe that I have
to think a bit more about it, since dropping the item identifiers of
the included file may not work if A includes B and B includes C.

Best regards,
Lars


-- 
Semagia
<http://www.semagia.com/>