I unfortunately have run into the same problem with my own product. I'm very interested in knowing as well how this should be interpretted.

On 2013-10-02 19:29, yoshito_umaoka@us.ibm.com wrote:
When I tried to import tzdata2013g into our project, I realized 2013g Morocco rule change revealed our tooling problem.

Our tool tries to extract a set of rules to be applied beyond year 2038, and expecting either no "max" rule or a pair of "max" rules. However, Africa/Casablanca with 2013g update introduced a case not fall into above - only a single "max" rule.

Of course, tzcode works fine and it's our tooling issue. My problem is how we should interpret the rule.

Below is a part of Morocco rules:

Rule        Morocco        2013        only        -        Jul          7         3:00        0        -
Rule        Morocco        2013        only        -        Aug         10         2:00        1:00        S
Rule        Morocco        2013        2035        -        Oct         lastSun 3:00        0        -
Rule        Morocco        2014        2022        -        Mar         lastSun 2:00        1:00        S
Rule        Morocco        2014        only        -        Jun         29         3:00        0        -
Rule        Morocco        2014        only        -        Jul         29         2:00        1:00        S
Rule        Morocco        2015        only        -        Jun         18         3:00        0        -
Rule        Morocco        2015        only        -        Jul         18         2:00        1:00        S
Rule        Morocco        2016        only        -        Jun          7         3:00        0        -
Rule        Morocco        2016        only        -        Jul          7         2:00        1:00        S
Rule        Morocco        2017        only        -        May         27         3:00        0        -
Rule        Morocco        2017        only        -        Jun         26         2:00        1:00        S
Rule        Morocco        2018        only        -        May         16         3:00        0        -
Rule        Morocco        2018        only        -        Jun         15         2:00        1:00        S
Rule        Morocco        2019        only        -        May          6         3:00        0        -
Rule        Morocco        2019        only        -        Jun          5         2:00        1:00        S
Rule        Morocco        2020        only        -        Apr         24         3:00        0        -
Rule        Morocco        2020        only        -        May         24         2:00        1:00        S
Rule        Morocco        2021        only        -        Apr         13         3:00        0        -
Rule        Morocco        2021        only        -        May         13         2:00        1:00        S
Rule        Morocco        2022        only        -        Apr          3         3:00        0        -
Rule        Morocco        2022        only        -        May          3         2:00        1:00        S
Rule        Morocco        2023        only        -        Apr         22         2:00        1:00        S
Rule        Morocco        2024        only        -        Apr         10         2:00        1:00        S
Rule        Morocco        2025        only        -        Mar         31         2:00        1:00        S
Rule        Morocco        2026        max        -        Mar         lastSun 2:00        1:00        S
Rule        Morocco        2036        only        -        Oct         21         3:00        0        -
Rule        Morocco        2037        only        -        Oct         11         3:00        0        -


If I read above, the rule applicable 2038 and beyond is only

> Rule        Morocco        2026        max        -        Mar         lastSun 2:00        1:00        S

Therefore, if I read this as is, Africa/Casablanca will be permanent daylight saving time after 2038-03-28.

I know the tz database is trying to cover dates up to 2037 (max signed integer seconds). Also, rules used in far future are not really useful. But users like us really need to make a certain assumption for far future dates.

I'm actually not 100% sure if we can still use the tz database for year 2038 and beyond, or if the tz database has no support / desire for make it work beyond 2038. My comments below is based on the assumption that the tz database just set 2038 as the threshold for rule coverage, but not trying to introduce the hard barrier.

One thing I'm not comfortable is that the rules above will result permanent daylight saving time. If the max line rule were written like below -

Rule        Morocco        2026        2037        -        Mar         lastSun 2:00        1:00        S

it makes better sense to me. The rule above indicates the last transition is on 2037-10-11, then it will stay in standard time beyond 2038. This is perfectly equivalent to what 2013g defined until 2038. Only the difference is 2038-03-28 (out of signed 32bit integer seconds) and beyond.

I would like to ask the tz database coordinators to select rule line more friendly for users using the database for 2038 and beyond.

Thanks,
Yoshito 


--