[tz] Zones names - stability, translation, groupings, CLDR, find-ability

Tobias Conradi tobias.conradi at gmail.com
Wed May 16 09:35:39 UTC 2012

was: [tz] Propose to improve where to find rules for "Indian/Kerguelen"

On Wed, May 16, 2012 at 8:30 AM, Mark Davis ☕ <mark at macchiato.com> wrote:
> I firmly agree. Stability is far more important than fiddling with the
> identifier names.
It's is documented practice of the tz project to "fiddle" with the
"identifier names", if fiddling refers to changing them, and
"identifier names" refers to zone names.

Also different defining location are contained in one link in the
backward file, e.g.
# @(#)backward	8.11
# This file provides links between current names for time zones
# and their old names.
Link	Africa/Bamako Africa/Timbuktu
Timbuktu is not an old name for the location Bamako.

> After all, the identifier names are really not for human consumption;
At least the tz project claims to attempt them to be:
"The time zone rule file naming conventions attempt to strike a balance
among the following goals:
 * Indicate to humans as to where that region is.  This simplifes use."

> they
> need translation for anything but English,
America/Argentina/Buenos_Aires does not look as if it would "need"
translation "for" Spanish.

> And if people want to have
> different groupings, that is easy to
> do.
Easy is relative. Even more "easy" would it be if there would be less
desire to have different groupings. To do so the IANA tz project could
follow international standards instead of using the current
self-invented world partitioning system, e.g. "Australasia" along side
"Asia", Bermuda in "Atlantic", Greenland in "America".

> In CLDR, for example, we group according to the United Nations M.49
> standard for continents and subcontinents.
> (Cf http://unicode.org/repos/cldr-tmp/trunk/diff/supplemental/territory_containment_un_m_49.html)
At least a try to match zones to an international standard continent
model, but the document
1) is not monitorable by robots that respect robots.txt rules, e.g.
the one from https://www.changedetection.com.

Disallow: /repos/cldr-tmp/	# dynamic

2) is out of date - zones are missing

3) has a n:m relationship for some cells in "Country (Territory)":"Time Zone"

"Country (Territory)":
Canary Islands	[IC]
Ceuta and Melilla	[EA]
map to one cell in "Time Zone"

4) has a n:m relationship for continent:zone, e.g. all zones matched
to the "Subcontinent" element "Caribbean" are matched to three
elements from the "Continent" column, namely to Americas 019, Latin
America 419, and North America 003.

5) lists these seven "continents" as per the "Continent" column:
Africa	[002]
Americas	[019]
Asia	[142]
Europe	[150]
Latin America	[419]
North America	[003]
Oceania	[009]
which form a set which is not used in any continent-model. Nor does it
look very widespread to
- have Latin America as continent at all,
- omit in a full world model the continent Antarctica and to put the
Antarctic Islands into Oceania.

6) contains names in the column headers which are not documented at
http://unstats.un.org/unsd/methods/m49/m49regin.htm, e.g.

7) contains unspecified names in the introduction, e.g. "CLDR
supplements this table with the QO code for outlying areas". "QO

8) doesn't specify which set of names is used for zone matching, it
says: "As the last column, the timezone IDs for that country are

but "Asia/Kolkata" contained in zone.tab is not listed for India,
instead "Asia/Calcutta" is which is not listed in zone.tab.

Also the document doesn't name the last column close to what it claims
to list, e.g. "timezone ID", instead it names the column "Time Zone".

> And for programmers, grep or equivalent is fine.
This is only true for the subset of programmers for which grep or
equivalent is fine.

It is not true for the others.

Tobias Conradi
Rheinsberger Str. 18
10115 Berlin


More information about the tz mailing list