W3C home > Mailing lists > Public > www-xml-schema-comments@w3.org > July to September 2000

RE: XML Schema: enumeration 'value' subtyping

From: <Stefan.Keller@lt.admin.ch>
Date: Wed, 9 Aug 2000 11:18:03 +0200
Message-ID: <5143BF97E6DDD1119F6F00805FC78025A33409@mlts1.lt.admin.ch>
To: curt.arnold@hyprotech.com
Cc: www-xml-schema-comments@w3.org, xml-dev@xml.org
Dear Curt,

Thank you very much for your help! I think you presented an interesting
approach of taking complex (abstract data) types instead subtyping an
attribute type. 

As a newbie I admit having problems in reading XML schema definitions: I'm
used in modeling with UML static class diagrams and INTERLIS. In order to
fully understand your proposal I would need to see the same in terms of an
UML class diagram or some other 'higher level' language (if I had your
example at hand in some UML class editor file format, that would be great).

You also mention some advantages in your complexType approach (vs.
simpleType):
1. Opportunities for extension must be specified in the base document.
2. Applications that are only aware of the federal level don't get confused
by canton or county specific information (=extensions) they don't recognize.

I'd like to say, that the second point is also possible with our approach:
we use an INTERLIS-specific XML-Format application. 

Regarding the first point I am not sure what you exactly mean: Before all,
the strength of subtyping (or subclassing) is, that one has not to foresee
specification any possible subdivision. This is the real point: In a federal
infrastructure with subsidiarity the central governement can not know and
foresee any possible extension (this applies also for international
standardization bodies regarding national specialities). It is essential,
that opportunities for extension don't have to be specified in the base
document (above this we offer in INTERLIS GeoLanguage that the central
governement forbids extensibility of these simpleTypes, similar to what the
keyword FINAL in class definitions does). Is this also possible in your
approach? or do I really have to indicate explicitly for all (federal)
LandCoverTypes, that there is an opportunity for extension?

Yours,
-- Stefan
______________________________________________________________________
Stefan F. Keller                  INTERLIS.net: http://www.gis.ethz.ch
Eidg. Vermessungsdirektion          Directorate of cadastral surveying
Bundesamt fuer Landestopographie          Federal Office of Topography
Seftigenstrasse 264                           Phone (+41)-31 963 22 44
CH-3084 Wabern (Switzerland)                    Fax (+41)-31 963 22 97
E-Mail: mailto:stefan.keller@lt.admin.ch  Web: http://www.swisstopo.ch
______________________________________________________________________
- A conceptual model should be really conceptual... (unknown/RB/SFK) -
Received on Wednesday, 9 August 2000 05:22:11 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 14:49:53 UTC