[tz] CLDR UN/LOCODE usage

Tobias Conradi mail.2012 at tobiasconradi.com
Sun Apr 21 11:24:34 UTC 2013

Was: Re: [tz] Proposal to change Macquarie Island to be Australian territory
A reply to :http://mm.icann.org/pipermail/tz/2013-April/019135.html
First LOCODE message in the thread was at

On Sat, Apr 20, 2013 at 3:53 PM, Mark Davis ☕ <mark at macchiato.com> wrote:
> On Fri, Apr 19, 2013 at 12:35 PM, Tobias Conradi
> <mail.2012 at tobiasconradi.com> wrote:
>> On Fri, Apr 19, 2013 at 11:17 AM, Mark Davis ☕ <mark at macchiato.com> wrote:
>> >> Variable length and inconsistent country code usage:
>> >
>> > This is a misunderstanding.
>> Of the UN LOCODE system, on the side of CLDR designers.
>> > CLDR is specified to use 5 letter UN LOCODEs where they exist. Where
>> > they do
>> > not exist, it is specified to use a non-5-letter code, precisely so that
>> > they do not overlap with future UN LOCODEs.
>> That could have been easily achieved with 5-letter codes, as has been
>> shown at:
>> http://mm.icann.org/pipermail/tz/2012-May/017974.html and
>> http://mm.icann.org/pipermail/tz/2012-May/017980.html
>> If overlap prevention the goal, then there was misunderstanding of the
>> UN LOCODE system on the side of CLDR.
> The only way that there would be a problem for CLDR is if the UN LOCODEs
> could be other than 5 alphanumerics.
It may be true that there is no /problem for CLDR/ - but the claim was
that CLDR has "Variable length and inconsistent country code usage"

> According to
> http://www.unece.org/fileadmin/DAM/cefact/locode/unlocode_manual.pdf, the UN
> LOCODE would be 5 letters and possibly numbers (in positions 3-5). As long
> as any additional CLDR identifiers did not match that, there wouldn't be an
> overlap.
> So I don't understand your contention. Are you saying that UN LOCODEs, can
> in fact consist of other than 5 alphanumerics?
I don't say so, and AFAICS never said so.

>> > When the codes are not 5
>> > letters, the first two letters have no meaning.
>> Except where they have. And for some from the US, even the first 4
>> have a meaning:
>> debsngn: de = Germany
>> gldkshvn: gl = Greenland
>> mxstis: mx = Mexico
>> rukhndg: ru = Russia
>> ruunera: ru = Russia
>> usinvev: us = US (usin = Indiana)
>> usnavajo: us = US
>> usndcnt: us = US (usnd = North Dakota)
>> usndnsl: us = US (usnd = North Dakota)
> I should have said: When the codes are not 5 alphanumerics, the first two
> letters do not *necessarily* denote a country code.
That was my starting claim: "inconsistent country code usage"

>> > The codes are stablized, meaning that they will not change no matter
>> > what
>> > changes happen in the base codes. So if Hawaii leaves the US and joins
>> > Canada as a new province, "ushnl" would not change in CLDR even if the
>> > UN
>> > LOCODE changes to "cahnl" or something else.
>> And combining this with "When the codes are not 5 letters, the first
>> two letters have no meaning.", assuming that, if they have 5 letters
>> they /may/ have a meaning, leads to the fact that the meaning in CLDR
>> would be different from the meaning in the UN LOCODE system.
> I don't understand what you are saying: that would only be an issue if there
> were overlaps.
What don't you understand in : "the meaning in CLDR would be different
from the meaning in the UN LOCODE system"?

The meaning of the first two letters in the UN/LOCODE system, except
for private range codes, is to show the current valid ISO 3166-1
alpha-2 country code for the country the location identified by the
UN/LOCODE is located in.

>> A statement on why a country relation was not seen suitable for time
>> zone identifiers can be found at:
>> ftp://ftp.iana.org/tz/code/Theory
>> "Be robust in the presence of political changes."
> That's fine for TZ codes. As far as CLDR is concerned, however, the main
> goal is to have an unambiguous, stable ID that is from 3-8 ASCII
> alphanumerics long and maps to TZ codes. The 3-8 ASCII limitation is imposed
> by BCP47, so we could not use the TZ codes unmodified. We could have chosen
> to simply number them, but decided that the UN LOCODEs would be a bit
> simpler to deal with.

A "3-8 ASCII limitation [...] imposed by BCP47" would have allowed for
fixed length 5-digit codes.

As emailed here:
"It is common for identifiers created by ISO to have fixed length, e.g.
language codes."

But CLDR for an unstated reason did choose variable length.

If one lists them in a table, the column containing them is
unnecessarily wider. If one puts them as labels on a map, they are
harder to fit.

In both cases, the non-5-character codes draw extra attention by the
reader to them.

Maybe all new assignments can be restricted to 5-character codes?

Tobias Conradi
Rheinsberger Str. 18
10115 Berlin


More information about the tz mailing list