[m-rev.] calendar module proposal

[Ian MacLarty]

On Tue, Jan 27, 2009 at 5:54 PM, Julien Fischer
<juliensf at csse.unimelb.edu.au> wrote:
>
> On Tue, 27 Jan 2009, Ian MacLarty wrote:
>
>> On Tue, Jan 27, 2009 at 04:38:52PM +1100, Ralph Becket wrote:
>>>
>>> This looks good.
>>>
>>> Ian MacLarty, Tuesday, 27 January 2009:
>>>>
>>>> Hi,
>>>>
>>>> Here's a proposal for a calendar module for the standard library.
>>>> Could interested people take a look and give me some feedback?
>>>> Since this module only implements the Gregorian calendar, it might not
>>>> be suitable for working with dates prior to 1582.  Perhaps it should be
>>>> renamed to gregorian_calendar or gcalendar?
>>>>
>>>> Ian.
>>>>
>>>> Index: library/calendar.m
>>>> ===================================================================
>>>> RCS file: library/calendar.m
>>>> diff -N library/calendar.m
>>>> --- /dev/null   1 Jan 1970 00:00:00 -0000
>>>> +++ library/calendar.m  27 Jan 2009 03:51:25 -0000
>>>> @@ -0,0 +1,961 @@
>>>>
>>>> +%-----------------------------------------------------------------------------%
>>>> +% vim: ft=mercury ts=4 sw=4 et wm=0 tw=0
>>>>
>>>> +%-----------------------------------------------------------------------------%
>>>> +% Copyright (C) 2009 The University of Melbourne.
>>>> +% This file may only be copied under the terms of the GNU Library
>>>> General
>>>> +% Public License - see the file COPYING.LIB in the Mercury
>>>> distribution.
>>>>
>>>> +%-----------------------------------------------------------------------------%
>>>> +%
>>>> +% File: calendar.m.
>>>> +% Main authors: maclarty
>>>> +% Stability: low.
>>>> +%
>>>> +% Proleptic Gregorian calendar utilities.
>>>
>>> Can prolepsy be cured with antibiotics?
>>>
>>
>> No, but it might be cured by implementing a hybrid Julian/Gregorian
>> calendar.  Such a calendar would use the Julian system before a given
>> crossover date and the Gregorian system after this date.  The crossover
>> date would need to be changable, because different countries started
>> using the Gregorian calendar at different times.  This cure would have
>> the side-effects of a more complicated interface and implementation.
>
> Premsuably the latter is exactly why that was what was done in the
> Java API!  (See:
> http://java.sun.com/javase/6/docs/api/java/util/GregorianCalendar.html>).
>
> Is it just matter of (optionally) allowing the crossover date to be set
> when creating dates and durations?
>

No, I don't think so.  Firstly I don't think the crossover date is
relevant for durations.  Secondly, you can't keep the crossover as
part of the date value, because then you have problems with
comparison.  If I have two dates with different crossover dates, but
that represent the same point in time (for example 9-10-1977 with
crossover 1-1-1500 and 9-10-1977 with crossover 1-1-1600) then how
should these dates compare?  They can't compare as equal, because if I
add a duration of negative 182500 days to each I'll get different
results.  Intuitively though they should compare as equal because they
represent the same point in time.

To avoid this problem we could create a separate calendar value which
we pass as an extra argument to the various calendar predicates.  The
calendar value would represent either a full Julian calendar, a full
Gregorian calendar, or a hybrid calendar with a crossover date.  Date
values would then use some canonical internal representation so that
the same points in time would compare as equal.

Ian.
--------------------------------------------------------------------------
mercury-reviews mailing list
Post messages to:       mercury-reviews at csse.unimelb.edu.au
Administrative Queries: owner-mercury-reviews at csse.unimelb.edu.au
Subscriptions:          mercury-reviews-request at csse.unimelb.edu.au
--------------------------------------------------------------------------