[tz] Urumqi timezone.

Guy Harris guy at alum.mit.edu
Thu Apr 13 19:52:00 UTC 2017

On Apr 12, 2017, at 10:17 PM, gfb hjjhjh <c933103 at gmail.com> wrote:

> ​But a problem with the current approach is that when some products opt to use geolocation-based timezone selection, users of UTC+8 timezone in the area that are not aware of the existence of UTC+6 timezone could be confused by why their gadgets would automatically get switched into UTC+6, as exampled by public reaction after one such iOS update. Ultimately that would make most manufacturer default the timezone for Xinjiang to UTC+8 instead since UTC+6-users are more aware of the situation and thus less likely to be confused.

If a vendor opts to use geolocation-based timezone selection, then, at least as I understand the issue described here, they will need to handle Urumqi specially, because knowing the *location* is insufficient to know the *time zone offset and rules* - some people there use the standard Chinese rules and some people there use the Urumqi rules.

That's (currently) outside the scope of the tzdb, as is providing shapefiles or whatever for the regions to which particular tzdb zones apply.

Any such collection of shapefiles-and-zones should support having *more than one* zone for a region and, for the region to which Xinjiang belongs, provide both Asia/Shanghai and Asia/Urumqi.  Any code using such a collection should do something appropriate, e.g. letting the user choose which one to use *and* remember that choice (so that, for example, if the user of some product has selected "Urumqi time", when a tzdb update is processed, the code doesn't switch to "regular China time").

If, with a tzdb update, iOS chose to remove /etc/localtime and link it to /usr/share/zoneinfo/Asia/Shanghai, even though it had been previously linked to /usr/share/zoneinfo/Asia/Urumqi, that's an iOS bug, and Apple should fix it.  (And the same applies if they happen to have a separate mechanism for specifying the current time zone in ICU or Foundation.)  It's not our problem and not our fault.

More information about the tz mailing list