[threeten-dev] Localized GMT offset

[Stephen Colebourne]

The prefixes in the CLDR "localized GMT offset" have nothing to do
with the ZoneId and its prefixes. Nothing at all. (Sorry, but I think
you've got confused between what CLDR is specifying and our ZoneId)

The CLDR localized GMT offset is only concerned with outputting a
ZoneOffset, in whatever style the localized data defines (typically a
prefix or suffix).

I can understand how it might be easy to confuse our ZoneId offset
prefixes with CLDR text that looks similar, but the two are very
different. See the two links I sent to see how the output is based
solely on the offset and localization data.

To get the CLDR data (which doesn't look especially complete) would be
an additional task for Oracle I suspect. Especially, as you'd have to
parse a format like "XXX {0}" and substitute the "{0}" for the
hour/minute part.

Leaving this for a later release would be perfectly reasonable option.
Stephen

On 22 March 2013 22:54, Xueming Shen <xueming.shen at oracle.com> wrote:
> The difficulty here is that those prefixes are not part of the JSR310 offset
> "zone" name,
> but regional zone name. And now we are talking about they are actually part
> of the
> offset formats, when if "localized". This is one the reason I suggested,
> maybe, just
> maybe those prefixes should be part of the offset "zone" names during the
> conf call.
>
> Anyway to treat it as a single "localizedOffset" form might be an
> alternative, just
> ignore above possible issue. But I don't think currently there is such
> "localization data"
> clearly defined in JDK and exists in JDK/CLDR data. Maybed we can skip this
> one for
> now and only add ZZZZZ for 8?
>
> Sherman
>
>
> On 03/22/2013 03:09 PM, Stephen Colebourne wrote:
>>
>> My interpretation of CLDR is that there is a single "localized offset"
>> format. So I would expect that the new method on the formatter builder
>> would be appendLocalizedOffset() where the format is derived entirely
>> from localization data (gmtOffset, gmtZeroOffset and hourFormat).
>> http://www.unicode.org/reports/tr35/tr35-dates.html#Primary_Zones
>>
>> http://www.unicode.org/cldr/charts/by_type/patterns.timezones.html#c394dccbd6c2e2d
>>
>> The GMT/UTC/UT prefixes are ZoneId forms, not directly related to
>> offset printing and not really relevant here
>>
>> Stephen
>>
>>
>> On 22 March 2013 20:22, Xueming Shen<xueming.shen at oracle.com>  wrote:
>>>
>>> Hi,
>>>
>>> I have not tried the "localized" resource for GMT yet (Masayoshi
>>> suggested
>>> there are such l10n resources in JDK, but few in CLDR data...), just
>>> wonder
>>> if this is the approach we want to go with to be"CLDR compatible" for
>>> zone
>>> pattern letters Z[4, 5] and O[1, 4]
>>>
>>> http://cr.openjdk.java.net/~sherman/jdk8_threeten/fmtOffsetPrefixed/
>>>
>>> -Sherman
>
>