[tz] Time Zone Database

[Paul Eggert]

On 3/16/23 01:26, EL MAAYATI Afaf wrote:

> (https://www.iana.org/time-zones <https://www.iana.org/time-zones>), for 
> Africa region, and especially for Morocco, it’s mentioned that a 
> program will be running to implement the transition dates.

Yes, I wrote that program to guess what Moroccan authorities will do in 
the future. Unfortunately, I don't know how to predict what Moroccan 
authorities will do, atop the underlying well-known uncertainty about 
when Ramadan will occur.

The program I wrote was conservative: when it wasn't absolutely sure 
that Ramadan would end by a certain time, it delayed the spring-forward 
transition by another week. This year Moroccan authorities did not delay 
similarly - i.e., they were less conservative than I was. Possibly this 
is because their astronomical calculations are more precise than mine, 
so that they can afford to be more aggressive in their predictions.

I worked around the 2023 problem by using a more-aggressive calculation 
as described here:

https://mm.icann.org/pipermail/tz/2023-March/032709.html

and this fix should be in the next tzdb release, which should be 
published soon.

This patch affects spring-forward predictions for 2023, 2031, 2038, 
2046, etc. However, it's likely that in the future the tzdb tables will 
need to be corrected again, as Moroccan authorities' calculation method 
evidently differs from the one published in tzdb.

If someone could let us know exactly how the Moroccan authorities 
calculate DST start and end time, we could presumably fix the tzdb 
calculations to match. This should avoid the need for last-minute 
corrections and so would make Morocco's clocks more reliable.