RE: [topicmaps-comment] referring to a topic from outside a TM

["Bandholtz, Thomas" <thomas.bandholtz@koeln.sema.slb.com>]

Title: RE: [topicmaps-comment] referring to a topic from outside a TM

> -----Original Message-----
> From: Thomas B. Passin
...
> Right- xlink uses XPointers.  However, the form xxx#yyy is an xpointer
> shortcut that means #xpointer(id(yyy)), as long as "yyy"
> actually is the
> value of an attribute of type ID.  Thus the use of
> xlink:href="#yyy" in a
> topicRef is a perfectly normal and correct use of
> xlink/xpointer, and refers
> to an ID in the same document. 

Right! I finally missed that yesterday. Sorry.

> Of course, without a DTD or schema, the processor cannot know
> whether a
> given attribute is of type ID or not.  Some xlink processors
> assume that an
> attribute named "id" is really an ID and treat it as such,
> some do not.

> (similar in : From: Geir Ove Grønmo)

ontopia.xtm should contain a namespace declaration.
XTM DTD contains namespaces as fixed attributs -
xmlns           CDATA     #FIXED 'http://www.topicmaps.org/xtm/1.0/'
xmlns:xlink     CDATA     #FIXED 'http://www.w3.org/1999/xlink'
xml:base        CDATA     #IMPLIED
- but ontopia.xtm does not refer to this at all.
- Secondly, http://www.topicmaps.org/xtm/1.0/ is a valid namespace name - but not a valid DTD location. BTW - is there a well defined, persistent location? (good URIs never change)

> You said
>
> "Remember there is also a third valid syntax (besides
> document-anchors and
> xpointer): a HTTP GET request like
>
>  http://psi.ontopia.net/ontopia/ontopia?id=topicmap"
>
> That requires a server to understand what is denoted by the
> query fragment,
> and we have no standard for that.  The server is free to
> retrieve whatever
> it chooses to.

Right. When you use XPointer like above you will have to maintain the named file and make sure it contains the named ID - what's the difference- The file just needs to contain _any_ element with whatsoever name that happens to have ID="topicmap" but containing "whatever  it chooses to".

Beeing bound to files is a legacy restriction. Topic Maps (even PS-collections) will be very large, so they will be maintained in databases, not in Gigabytes of XTM-files.

What about a little Java Servlet, CGI, or pl that returns a valid XTM encoding of the topic with the given ID?

This will not position to a certain point in a probably very large document, but it will return exactly one filtered <topic>. range in a filtered <topicMap>. This should be a legal solution.

If you supply one URL for topic-by-id retrieval and one for associations-by-topicid, this will give access to everything (ok, add mergeMap).

Nobody would have to merge copies of your map into his own - this can be merged "virtually" and distributed.

Thomas Bandholtz
XML Competence Center
SchlumbergerSema
Sema GmbH
Kaltenbornweg 3
D50679 Köln/Cologne
++49 (0)221 8299 264