On 2021-05-10 08:38, Paul Eggert wrote:
The idea is to move Zones into 'backward' if they're identical to some other Zone after 1970.
If these rules really hold for backzone (are they listed anywhere?), then America/Curacao also should go to backzone, and one of Africa/Johannesburg and Africa/Maputo and three of Pacific/Majuro, Pacific/Tarawa, Pacific/Funafuti, Pacific/Mata-Utu should go to backzone. By the way, do these rules imply that Pacific/Yap with the data as of 2005k belongs to backzone? A remark on backward compatibility: The proposed change is not conservative in the sense that information that could automatically be extracted from previous versions is no longer available in this way. To wit: It is no longer possible to automatically find geographical coordinates for the tzdb timezones described in backzone. In previous versions, the name of a tzdb timezone in backzone could also be found in zone.tab together with its country code and the coordinates for its location (except for backzone/Asia/Hanoi). After the change, there is no systematic way to find the right line in zone.tab with the coordinates, even if one supposes that one can automatically deduce the ISO country code for the name of a tzdb timezone in backzone (which is quite tricky and not so simple as in previous versions). Only the comment entry in a line of zone.tab (such as "Vietnam (south)") gives some indication on the name of the timezone (eg, Asia/Ho_Chi_Minh) whose geographic coordinates are contained in that line. I am aware that zone.tab is deprecated, so that one could say that the previously possible method was not guaranteed to work across versions -- but then the locations for backzone timezones could be omitted or put somewhere where they can be found automatically. Michael Deckers.