We have actually instituted an alternative mechanism called 'metazones', which are somewhat like what you describe. They basically coalesce zones that have the same behavior in modern time. However, they have some further restrictions, and we only put this in recently thus haven't gathered enough data to be useful yet.

Mark

On 7/10/07, Chuck Soper <chucks2@veladg.com> wrote:

I've been on the tz list for several years. I have tried to follow the Time Zone Localizations topic for CLDR.

I do not understand why translations would be supplied for timezone IDs (tzIDs). Many tzIDs refer to the same time zone name. For example, I believe that Argentina has 2 time zone names, yet it has 10 tzIDs to handle DST rules for various states. Does this mean that 20 localized names would be required for Argentina? Does this mean that a new tzID would not be localized?

Looking at 381 tzIDs of a tzData version (from 2006), I multiply it by two (for std and dst names) to obtain 762 time zone names required. By removing names of tzIDs that do not recognize DST and removing tzIDs that refer to the same time name (e.g. many tzIDs refer to Central European Time), I was to reduce total time zone names from 762 to 212.

Can over 500 names can be removed from the translation list? Multiplying 500 potentially unneeded names by the number of translations would result in a large reduction of translation efforts.

My approach is dependent on having a stable list of time zone names. As far as I know, such a list does not exist. If such a list existed, I would envision tzIDs being mapped to the 'time zone name' list.

Chuck

At 10:44 AM -0700 7/10/07, Mark Davis wrote:

The Unicode CLDR project (http://unicode.org/cldr/) does supply translations for timezone IDs. There are a few caveats.

1.      The timezone database really has equivalence classes of IDs. One of these can be used as a representative for any in the equivalence class. It is the zone.tab file that contains such IDs. CLDR started by using that file, but unfortunately it is not stable (different equivalent IDs can be substituted at any time). So what we do is use as the representative the one that historically the first one used in any zone.tab file (after CLDR started).

2.      We allow, but do not encourage, translation of zones that are the only zone in a country. For that we use the country name. This cuts down very substantially on the number of translations needed. That is, you would see the equivalent of "Italy", and "United States (Los Angeles)" -- only in the latter case do we need translations for the cities.

3.      Translators can optionally add other variations: daylight (summer) time, standard (winter) time, and generic time, both abbreviated and long.

4.      These choices percolate out to clients of CLDR: Google, IBM, Apple, Adobe, and many others.

Mark

On 7/10/07, Olson, Arthur David (NIH/NCI) [E] <olsona@dc37a.nci.nih.gov> wrote:

I'm forwarding this message from Vincent Untz, who is not on the time zone mailing list.

Those of you who are on the time zone mailing list should direct replies appropriately.

                                --ado

-----Original Message-----
From: Vincent Untz [mailto: vuntz@gnome.org]
Sent: Tuesday, July 10, 2007 1:24 PM
To: tz@lecserver.nci.nih.gov
Subject: Localization of timezones from zone.tab

Hi,

[please keep me cc'ed since I'm not subscribed to this list]

I know that, at least in GNOME, there are now three places where we

parse zone.tab to get a list of timezones supported by the OS. I suppose
other projects are also doing this. This list is then presented to the
user to let him choose the timezone.

The problem here is that we let the user choose strings which look like
"Antarctica/South_Pole". This is not really good for
non-english-speaking people ;-)

Of course, we can add all the timezones to our list of strings to
translate, but this means all projects needing to do so will duplicate
this work and the translations.

I'd like the tz database to ship translations in po files. This would
imply the following:

+ create a small script to generate a POT file from zone.tab (easy)
+ submit the POT file to the translation project [1] (or any other

   place that helps with translation)

+ add the po files for translation to the tz database
+ choose a gettext domain

(Of course, I'm quite probably forgetting about a step :-))

I've seen that this topic has been discussed before [2], but the
proposition there was really more ambitious, so I'm hoping a simple
approach would be welcomed.

What do you think?

Thanks,

[1] http://translationproject.org/
[2] http://osdir.com/ml/time.tz/2004-09/msg00000.html

Vincent

--
Les gens heureux ne sont pas pressés.

--
Mark

--
Mark