Re: [Pkg-isocodes-devel] Timezones?
Clytie Siddall a écrit :
> Thanks very much for your helpful and enthusiastic answers. :)
>
> On 21/01/2008, at 9:20 PM, Aurelien Jarno wrote:
>
>> On Sun, Jan 20, 2008 at 11:20:01AM +0100, Christian Perrier wrote:
>>> (if answering, please keep CC and put "X-PTS-Approved: yes" in
>>> headers
>>> so that tzdata@packages.debian.org gets the mail)
>>>
>>> Quoting Clytie Siddall (clytie@riverland.net.au):
>>>> Hi guys :)
>>>>
>>>> I'm in the middle of one of my running efforts to get developers to
>>>> use iso-codes as a plugin/library for their software, so they won't
>>>> have long lists of language names, country names, currency names
>>>> etc.
>>>> in their programs which we have to translate over and over again,
>>>> and
>>>> which they invariably fail to keep accurate or current.
>>>>
>>>> In this case, it's Evolution (GNOME).
>>>>
>>>> While paging through the huge Evolution translation file, I note
>>>> that,
>>>> in addition to a lo-o-ng list of language names, and an equally long
>>>> list of time/date data they could get from the locale, Evo lists
>>>> timezones. e.g.
>>>>
>>>> #.
>>>> #.* These are the timezone names from the Olson timezone data.
>>>> #.* We only place them here so gettext picks them up for
>>>> translation.
>>>> #.* Don't include in any C files.
>>>> #.
>>>> #: ../calendar/zones.h:7
>>>> msgid "Africa/Abidjan"
>>>> msgstr "Châu Phi/Abidjan"
>>>>
>>>> Any chance iso-codes will take on timezones?
>>>
>>> Could be interesting. But not necessarily the way to go.
>>>
>>> However, that'd require some coordination with fellow people who
>>> (well) maintain such stuff. In Debian, the tzdata package is now well
>>> developed and maintained, for instance.
>>>
>>> I don't really know if timezones are normalized in some way. If they
>>> are, it could make sense to imagine moving this to iso-codes (so that
>>> it could benefit people outside Debian). But timezones data goes far
>>> beyond simple names: it includes information about the shift wrt UTC
>>> as well as Daylight Savings Time information.
>>>
>>>
>>> If timezones aren't normalized, I think it
>>> would maybe make more sense to keep this in a separate
>>> package/distribution.
>> tzdata now works on a Zone/City basis. I am not sure the iso-codes
>> also
>> define city names.
>>
>> Then I guess it is mainly a question for the translators. We don't
>> mind
>> if the names are translated by hand or using the data from iso-codes
>> as
>> long as we have a .po file to drop in debian/po.
>>
>
> From the translator POV, the great advantages of standardizing these
> lists via iso-codes are:
>
> 1. the original strings are comprehensive, accurate and correctly
> spelt (!)
>
> 2. we only have to translate them once, and maintain that single
> translation
I don't think it is very easy to change the names in tzdata (except if
they are wrong). They are standardized among various UNIX systems, so
the only way it can work is by importing the names from tzdata into
iso-codes. Also a patch to change the names will be difficult to
maintain wrt new upstream versions.
>From my point of view, this has to be done in the reverse way, that is
importing cities from tzdata into iso-codes, and then using the
translation from iso-codes in tzdata.
> You might be amazed at the number of programs which still have mis-
> spelt, inaccurate and messy lists of language names, country names
> etc. And we have to translate them again, and again, and submit
> detailed bug reports about the inadequacies of the original strings.
>
> So I'm even keener to support iso-codes, including implementing other
> information that needs standardizing. ;)
>
> BTW, even if you haven't listed city names yet, and it would be
> impractible to list _all_ city names, the timezones are based on
> regions and sub-regions, so they are related to iso_3166-2. Perhaps
> the city names could be classified as part of a regional name.
>
> For example, as in:
>
> South Australia/Adelaide
>
> the timezone is the capital city tagged on to the state name.
>
> The above is my timezone (+0930, currently +0130), and it is often
> left out of supposed comprehensive timezone lists. ;)
>
This is not as easy. Timezones sometimes depends on the city, and can
differ from city to city within the same region. For example there are a
few cities which don't apply DST rules, while their region applies it.
Also tzdata provides timezone information for the current date, but also
in the past, back to early 20th century, when timezones where far less
standardized than now.
--
.''`. Aurelien Jarno | GPG: 1024D/F1BCDB73
: :' : Debian developer | Electrical Engineer
`. `' aurel32@debian.org | aurelien@aurel32.net
`- people.debian.org/~aurel32 | www.aurel32.net
Reply to: