[xtm-wg] Fixing language.xtm and country.xtm

[Lars Marius Garshol <larsga@garshol.priv.no>]

Now that it has been decided that TopicMaps.Org is going into OASIS I
think the first thing that needs to be done is to fix the problems
with the language.xtm and country.xtm PSI sets. These PSI sets are
very important (since they're used by many topic maps), and since
they're also part of XTM 1.0 I think it is highly important that they
work properly.

It has been well known for a while now that it needs fixing, and
nothing has been done (for valid reasons, I know), but I think the
time has now come when these PSI sets must be fixed. The longer we
wait the more incorrect topic maps will be created.

To help people understand the importance of this, here is a list of
the problems with these PSI sets as they are today:

 - both files have URIs http://www.topicmaps.org/xtm/1.0/foo.xtm
   and xml:base attributes "http://www.topicmaps.org/xtm/1.0/".

   This means that when a conforming XTM processor loads these topic
   maps, it will give the topic Afghanistan the source locator
   <URL: http://www.topicmaps.org/xtm/1.0/#AF >, since it must be
   relative to the base URI of the document element. (This, BTW,
   illustrates the need for a proper processing model. Only my
   proposal actually provides any guidance on what the correct
   behaviour is here.)

   This URI will fail to resolve to this element, though, since only
   the address <URL: http://www.topicmaps.org/xtm/1.0/country.xtm#AF >
   will actually find it. If you use this subject indicator and merge
   your topic map with country.xtm the topics will either fail to
   merge, or get two subject indicators...

   This also means that many people are bound to be confused, and give
   their topics subject indicators like
   <URL: http://www.topicmaps.org/xtm/1.0/country.xtm#AF > whereas
   others will get it 'right' and use
   <URL: http://www.topicmaps.org/xtm/1.0/#AF >. In other words, all
   these topics have two different subject indicators, and so when
   merging you have 50% chance of getting it right.


 - both PSI sets lack a topic that represents the type of these
   topics. This means that I can't unambiguously say that this topic
   is an instance of country, beacuse there is no country topic in
   country.xtm. This is actually a major problem, since I then risk
   having my topics end up with types 'Country' and 'country', for
   example. 


 - the topics in these PSI sets have the type 'country-name' and
   'language-code', which is clearly contrary to the intention of
   these PSI sets. This is related to the problem above, of course.


 - language.xtm only contains the two-letter codes for topcis, and it
   also only contains topics for the languages that have two-letter
   codes, but many languages only have three-letter codes, which means
   that the PSI set is incomplete.


 - the copyright statement refers to the XTM DTD, but surely it should
   only discuss the PSI sets themselves?


 - #lang-code has a variant name that IMHO is not a variant name at
   all.


 - languages.xtm is not up to date with the latest version of ISO 639;
   for example, the codes for Norwegian Bokmål and Nynorsk are missing.
   Cornish, introduced in August 1998, is also missing.
   <URL: http://lcweb.loc.gov/standards/iso639-2/codechanges.html >

 
 - the topics in these topic maps only have the English names for
   these topics, even though both standards also have French names for
   the topics. IMHO the French names should also be present.


In short, I think this needs to be fixed as soon as possible. I also
think that if the current editor doesn't have time to do this soon
then someone else should be given the job.

Reactions? Opinions? Comments?

--Lars M.


To Post a message, send it to:   xtm-wg@eGroups.com

To Unsubscribe, send a blank message to: xtm-wg-unsubscribe@eGroups.com 

Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/