Paul Hill <phill@myriad.com> writes:

But I believe the tzdata actually redundantly lists some entries. If I recall, the US territories in the Pacific (Guam?, American Samoa?) are listed in two files but with the same IDs.

There are many redundancies if you look at the "backward" file, since it contains backward-compatibility links for older names like US/Pacific that have been superseded by newer names like America/Los_Angeles. But if you omit the "backward" file, almost all redundancies are where two different locations have the same time zone history since 1970 (e.g. Europe/Rome and Europe/San_Marino). These redundancies are needed only to satisfy the rule that each country gets at least one location. There is only one exception to this general rule: Istanbul has the two entries Europe/Istanbul and Asia/Istanbul. I think this is OK, as Istanbul actually does span continents. I don't know of any current official dividing line between Europe and Asia, or between Asia and Africa for that matter, so I suppose there could be more points of controversy in the distant future as the dividing line encounters time zone splits. Maybe by then there will be a standard, or we'll use a different naming scheme. As for Cyprus, the tz database generally tries to avoid political and cultural questions, as they are distractions from its intended purpose. Culturally, Cyprus is more European than Asian (and has been so for most of its history); but if we went by culture, we'd have to do things like putting Australia under Europe and this wouldn't make much sense from a time zone point of view. Geographically, Cyprus is clearly in Asia, and this has long been recognized. For example, please see Herodotus, Histories, I.72.