[tz] Proposed patch - Theory change to reinstate reference to ISO-3166

[Clive D.W. Feather]

Tobias Conradi said:
> There are two standards containing the string ISO 3166-1, the one
> published the latest is
> ISO 3166-1:2006
> http://www.iso.org/iso/home/store/catalogue_tc/catalogue_detail.htm?csnumber=39719
> 
> The other is
> ISO 3166-1:1997
> http://www.iso.org/iso/home/store/catalogue_tc/catalogue_detail.htm?csnumber=24591
> 
> The ISO website also shows
> ISO 3166:1981
> ISO 3166:1988
> ISO 3166:1993

Those are all the same standard. They represent 5 different *revisions* of
the standard. Once 3166:1988 was published, 3166:1981 was no longer the
standard and should not be used. (Of course, ISO can't force people to
change.)

> ISO 3166-1:2006 includes three sets of codes.

No, contains *rules* for three sets of codes.

> There is no 1:1:1 correspondence between the sets. Not even 1:1
> between any pair of them.

The standards don't include a normative table of codes. That is, the tables
in those standards are *snapshots*.

The standard itself defines rules for maintaining the official list of
codes, what the codes should look like, and so on. The official list,
however, is maintained by the "maintenance agency" (who, IIRC, is DIN).
They can update the list whenever they want. Thus, for example, the list
was changed on:

1989-12-05   (removed BU, added MM)
1990-08-14  
1990-10-30  
1992-06-15
1992-08-28   (removed DD, changed DD from 280 to 276)
1992-08-30  
1993-06-15   (removed CS, added CZ and SK)
1993-06-18  
1993-07-12  
1993-07-16  
1993-07-22  
1993-07-23  
1993-07-28  
1996-04-03   (changed name of VA)
1997-07-14  
1999-10-01 
2002-02-01 
2002-11-15 
2003-07-23   (removed YU, added CS)

among other dates. (I've only given a sample of what the changes were for.)

> There are at least two pairs, namely DD/DE and YD/YE where there is
> not "at least one name per ISO-3166 code" but only one per pair.

When DD and DE were both in the list, they had separate names. DD is no
longer in the 3166-1 list, though it is in 3166-3 (removed codes).

> When looking at the locations or location names, it follows that
> "There should typically be at least one name per ISO-3166 code." was
> never applied.

I don't believe you've demonstrated that.

-- 
Clive D.W. Feather          | If you lie to the compiler,
Email: clive at davros.org     | it will get its revenge.
Web: http://www.davros.org  |   - Henry Spencer
Mobile: +44 7973 377646