On Wed, May 13, 2009 at 11:38 AM, Chip Davis <c...@aresti.com> wrote:
> Yes, that was the intention of the ANSI Standard Committee's Y2K work. It
> is not possible to convert from any of the ambiguous date formats. What
> should Date('U', 'Tuesday', 'W') return?
>
> Since the Julian format was defined to have only a two-digit year (yynnn)
> there is no way to define an accurate equivalent date. And yes, we
> discussed sliding windows and a number of other hacks and ultimately
> rejected them as dangerous. One could get the wrong answer and not know it.
> We like the 'astonishment factor' to stay as low as possible.
>
> Besides, as Bob has shown, there are several ways of accomplishing the
> conversion as long as you know the century the Julian date belongs to.
>
> -Chip-
>
However, date('S','04/05/09','U') returns '20090405' consistent with:
2. Input dates given in 2-digit year formats are interpreted as being within
a 100 year window as calculated by:
(current_year - 50) = low end of window
(current_year + 49) = high end of window
so it looks like the point is phasing out Julian dates.
>
> On 5/13/09 14:02 Edward M Martin said:
>
>> Hello Howard,
>>
>>
>> From z/VM 5.3 Help REXX DATE
>>
>>
>>
>>
>>> >--DATE(--.-----------------------.--.-------------.--)---------------><
>>>
>>
>> | (1)| '-| Group 1 |-'
>> '-output_date_format----'
>>
>> And then reference (1)
>>
>> Has
>>
>> NOTE:
>> (1) If the Century or Julian format is specified, then no other options
>>
>> are permitted. These two formats are provided for compatibility with
>>
>> programs written for releases prior to VM/ESA(R) version 2 release
>> 1.1. It is recommended that they not be used for new programs.
>>
>> It would seem that if you specify Julian format you have to use today’s
>> date.
>>
>>
>> Ed Martin
>>
>> Aultman Health Foundation
>>
>> 330-363-5050
>>
>> ext 35050
>>
>>
<snip>
