Re: [icu-support] DateTime formatting issues with pre-Gregorian dates

[Jamie D. <jam...@gm...>]

Done. http://bugs.icu-project.org/trac/ticket/12655

As an aside, trac doesn't like the + in the email I use to filter out ICU
issues from my normal mail.

-Jamie.

On 28 July 2016 at 17:22, Steven R. Loomis <sr...@ic...> wrote:

> Thanks. Would you mind filing a bug report?
>
> El [FECHA], "[NOMBRE]" <[DIRECCIÓN]> escribió:
>
> I've found what seems to be a workaround...
>
> UDate ToICUDate(const DateTime& DateTime)
> {
> //const int64 UNIXTimestamp = DateTime.ToUnixTimestamp();
> //return static_cast<double>(UNIXTimestamp) * U_MILLIS_PER_SECOND;
>
> UErrorCode ICUStatus = U_ZERO_ERROR;
>
> icu::GregorianCalendar GregCal(ICUStatus);
> GregCal.setTimeZone(icu::TimeZone::getUnknown());
>
> int32 Year, Month, Day;
> DateTime.GetDate(Year, Month, Day);
> GregCal.set(Year, Month - 1, Day, DateTime.GetHour(),
> DateTime.GetMinute(), DateTime.GetSecond());
>
> return GregCal.getTime(ICUStatus);
> }
>
> The commented out code at the top is how we used to convert to a UDate
> (via a timestamp), however if I instead marshal the time through
> a icu::GregorianCalendar instance, I get back the UDate that ICU is
> expecting and pre-Georgian dates seemingly format correctly (even without
> having to change the calendar used by the icu::DateFormat).
>
> I'm not terribly happy with this as a solution though, and I'd still like
> to know why the timestamp method is having issues as I don't believe our
> representation of time is that unusual.
>
> -Jamie.
>
> On 28 July 2016 at 00:49, Jamie Dale <jam...@gm...> wrote:
>
>> FWIW, before asking here I did try and search through the mailing list,
>> and this post from 2006 kept cropping up:
>> https://sourceforge.net/p/icu/mailman/message/12529106/
>>
>> This is strikingly similar to what I'm seeing (1st Jan 0001 is 2 days off
>> - I get 3rd Jan 0001), but given this is from 2006, surely it's been fixed
>> by now?
>>
>> -Jamie.
>>
>> On 27 July 2016 at 23:40, Jamie Dale <jam...@gm...> wrote:
>>
>>> Hey all,
>>>
>>> I'm investigating an issue where pre-Gregorian dates are failing to
>>> format correctly with icu::DateFormat.
>>>
>>> We're using ICU 53, and our internal concept of time is ticks of 0.1
>>> microseconds since 1st January 0001 - we convert this to a UNIX timestamp
>>> and multiply by U_MILLIS_PER_SECOND before passing it to ICU as a UDate.
>>>
>>> At first I thought this might be due to the Gregorian switch-over, so I
>>> tried constructing my own icu::GregorianCalendar, setting the Gregorian
>>> change date to 1st January 0001, and then letting the icu::DateFormat adopt
>>> that as its calendar, however it didn't seem to make any difference (my
>>> calendar reports U_USING_FALLBACK_WARNING when being constructed... I'm not
>>> sure why).
>>>
>>> I'm wondering if anyone else has seen this issue, or whether anyone has
>>> any advice?
>>>
>>> Thanks,
>>> Jamie.
>>>
>>
>>
> ------------------------------------------------------------------------------
> _______________________________________________ icu-support mailing list -
> icu...@li... To Un/Subscribe:
> https://lists.sourceforge.net/lists/listinfo/icu-support Archives/Project
> Info: http://site.icu-project.org/contacts
>
>
>
> ------------------------------------------------------------------------------
>
> _______________________________________________
> icu-support mailing list - icu...@li...
> To Un/Subscribe: https://lists.sourceforge.net/lists/listinfo/icu-support
> Archives/Project Info: http://site.icu-project.org/contacts
>
>