Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:103277
MIME-Version: 1.0
In-Reply-To: <ab53ddcd-3684-01b7-6218-82cc6b0db4c4@gmx.de>
References: <CAGyY9iQVAiFzC96eB51sczJuaZm1sAdJbaHLGXUFHmD94p-fDQ@mail.gmail.com>
 <ab53ddcd-3684-01b7-6218-82cc6b0db4c4@gmx.de>
Date: Fri, 28 Sep 2018 19:27:23 -0500
Message-ID: <CAGyY9iThZUKwT+q9wto1V3ZwDdAF294s_32VHhw8n37L0aUMWA@mail.gmail.com>
To: internals@lists.php.net
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Subject: Re: Proposal: Add tabular Islamic calendar to calendar conversion functions
From: weberk294@gmail.com (Kurt Weber)

Gah, I had my filter for this mailing list misconfigured and so I just
now saw this. So so so sorry.

I'll definitely do a RFC.  Regarding issue (c), I'm obviously not
familiar with internal PHP development processes, but is it usually
the case that "You have X similar thing" is an argument for doing Y?
Seems like there's no reason these decisions couldn't be made on an ad
hoc basis--tabular calender is one thing because it's entirely
self-contained, while other calendars would require access to an
external data source that may disappear at any time, etc.

On Sat, Sep 1, 2018 at 5:09 AM, Christoph M. Becker <cmbecker69@gmx.de> wro=
te:
> On 30.08.2018 at 22:33, Kurt Weber wrote:
>
>> The calendar conversion functions currently support (via Julian Day
>> Numbers as an intermediary) conversion between Gregorian, Julian,
>> Hebrew, and French Revolutionary calendars.  Conspicuously absent is
>> the Islamic calendar.  A comment in ext/calendar/sdncal.h seems to
>> suggest that the original author(s) intended to implement at some
>> point, but never got around to it.
>>
>> The Islamic calendar offers some complications because in most
>> places and communities that use it for religious purposes, the switch
>> from one month to the next is based on actually observing the moon in
>> the proper phase.  In theory the date of this change can be
>> calculated (and indeed the early Islamic surge in study of astronomy
>> was motivated by the wish to know about when they should start
>> looking), but tradition still demands that it be actually observed
>> before a change is made--the actual impact of this is that local sky
>> or weather conditions can sometimes interfere with the observation in
>> a particular jurisdiction, meaning that often it comes a day or so
>> later than when it otherwise would.
>>
>> Consequently, what I'm proposing is the Tabular Islamic Calendar,
>> which was specifically created to be predictable and calculable.  It
>> is of limited use for religious purposes (and documentation should
>> probably be clear about this), but it would be useful for, e.g.
>> people working with historical documents from Islamic communities or
>> groups (In fact, this is how I came to this problem--I am working on
>> a Ph.D. in Russian history, and as a side project I'm developing a
>> PHP-backed website to manage research notes and other information;
>> because I work on prerevolutionary Russia I'm very familiar with the
>> issue of needing to convert between calendar systems (in my case,
>> Julian and Gregorian), and my colleagues working on Islamic history
>> often have similar issues.).
>>
>> Tabular Islamic calendar is not ideal, but it seems like the only
>> realistic option for automated conversion.
>>
>> As far as implementation goes, I could do it.  I've already
>> implemented conversion one way, before I stopped work in case there
>> were issues that come up in this discussion that I hadn't foreseen
>> (unfortunately, that happened roughly at the time the mailing list
>> server seems to have died, so it's been paused for a couple of
>> weeks). So all I'd have to do is the other direction, and it'd be
>> ready to test.
>
> Thanks!  I'm not opposed to this suggestion.  There is the suspended
> feature request #27453[1], and I agree that even a tabular Islamic
> calendar would be much more useful than the French Revolunationary calend=
ar.
>
> Presently, I see three counter-arguments:
>
>   (a) already supported by ext/intl
>   (b) maintenance burden
>   (c) opening up a can of worms
>
> Ad (a): Islamic calendars are already supported by the
> intl extension[2].  Even though ext/intl supports an Islamic calendar,
> that doesn't prohibit that ext/calendar also does, though, since they
> are independent extensions (cf. ext/mbstring, ext/iconv, ext/recode).
>
> Ad (b): if nobody else is willing to step up as maintainer of
> ext/calendar[3], I'd be willing to do so.  I've already did some
> maintenance[4] so far.
>
> Ad (c): this is somewhat of a problem.  I can easily imagine folks
> requesting support for specific Islamic calendars which would require
> (up-to-date) data, what could easily blow ext/calendar out of
> proportion.  Also I can imagine people coming up with =E2=80=9Cthe FOO ca=
lendar
> is used more widely than the tabular Islamic calendar, so it should be
> supported as well=E2=80=9D argument.
>
> Especially due to (c) I suggest to go through the RFC process[5],
> whereby the RFC could also set some precendence/rules for potentially
> further calendar systems (i.e. what ext/calendar may support, and what no=
t).
>
> [1] <https://bugs.php.net/bug.php?id=3D27453>
> [2]
> <http://icu-project.org/apiref/icu4j/com/ibm/icu/util/IslamicCalendar.htm=
l>
> [3]
> <https://github.com/php/php-src/blob/7956722cfd96fdc244e9ed3dd13e162094be=
09cd/EXTENSIONS#L257>
> [4]
> <https://bugs.php.net/search.php?cmd=3Ddisplay&package_name[]=3DCalendar+=
related&direction=3DDESC&limit=3D30&assign=3Dcmb&status=3DAll&reorder_by=3D=
ts2>
> [5] <https://wiki.php.net/rfc/howto>
>
> --
> Christoph M. Becker