Re: glib time functions

[Ali Abdin <ALIABDIN aucegypt edu>]

On Thu, 30 Nov 2000, Robert Brady wrote:

> On Wed, 29 Nov 2000, Ali Abdin wrote:
> 
> > Also, some years, the muslim world is "divided" on the ramadan times. By
> > definition, the Islamic calendar can never be 100% accurate, but it can be
> > damn close (as proved by the lisp algorithm that was devised).
> 
> Are there a countable number of places that observations are done in? Are
> the differences at all predictable?

Each country does its own observation. If a country does not have 
"Islamic scholars" to do this observation, you base it on the closest 
muslim country that does (or something like that).

> What happens when it's wrong; presumably we want to correct this ASAP, so
> it should be easily replacable.
> 
> /me imagines NTP servers with an extension... :)

Yes, I had though of this actually. But an NTP server will not be a 
solution for "offline" applications and such.
 
I think that there should be an API function that allows you to take into 
account ramadan time. The application should (through an NTP server) 
what the ramadan is (like I said, G_RAMADAN_EARLY, G_RAMADAN_NORMAL, 
G_RAMADAN_LATE).

So an app would look up ramadan time on a "custom" NTP server and say, oh 
- we're early, then it would set the appropriate API function (which glib 
will now use to compute the dates (it will only affect dates that come 
/after/ the 9th month (i.e. it will only affect 3 months).

> > I have made one premise though (which I believe is safe to make) and that is
> > we don't determine new days based on sunset/sunrise (too difficult, not
> > essential, not used in the Islamic world anyway). 
> 
> That'd be very interesting, but would require glib knowing your latitude,
> longitude, altitude, whether there are mountains, and probably some other
> things.

Somebody was working on a libastro library that would actually "handle" 
this - You provide the timezone I think and you can get sunset/sunrise 
times.

Remember this is for /Date/ functions - I do not think time should be 
taken into account, because in the Islamic world nobody bases their time 
on Sunset/Sunrise (well, only for prayers). We should just use the 24 hr 
scheme.

> > If people want this for glib, I can try to work up a patch (if you tell me
> > what kind of functions you want (i.e. do you want just the conversion
> > functions? Should we add other things? etc.))
> 
> I'm going to do an initial patch for a few more simple calendars.
> (Gregorian, Julian, Persian, and probably some interesting proposed reform
> ones, and of course the Shire Reckoning (used by hobbits c.a. 3000 TA,
> which is an interesting one due to it having days that aren't weekdays.))

I'd do a patch too, but I got exams and finals coming up :(