[topicmapmail] occurrence abuse ? Was: [geolang-comment] Firstproposals for ISO 639 and 3166 available

[Anthony B. Coates]

** Reply to message from Piotr Kaminski <piotr@ideanest.com> on Tue, 20 Aug
2002 15:35:51 -0700

> The above is slightly tongue-in-cheek... but only slightly.  If we don't
> want occurrences to become generic binary relationships with no explicit
> member roles, then we ought to tighten up the definition of occurrence.
> (Or, perhaps, get rid of occurrences altogether.)

I tend to view occurrences as anonymous topics referring to particular data
values (some of which may be URIs).  They are allowed to be anonymous (i.e.
unnamed) because the implicit topic is uniquely identified by the particular
data value.  You could get rid of them in favour of a more heavyweight approach
using a full topic for every data value that occurs in a topic map, but I don't
think non-theoreticians would find that helpful, they would just find it onerous.

I guess in this view, occurrences should contain only simple data types,
without structure.  But what is structure?  In truth, it depends on what
software library you are using.  If you have support for ISO dates/times in your
software library, then a date/time is a simple type.  If you don't, it is a
piece of structured text that you have to parse and process.  So, there will
never be a black-and-white way of deciding what should be considered a "simple"
datatype for use as occurrence data.  It will at best be a guiding principle,
just as it is for XML attribute values.

	Cheers,
		Tony.
====
Anthony B. Coates, Financial XML Specialist
mailto:abcoates@TheOffice.net
MDDL Editor (Market Data Definition Language)
http://www.mddl.org/