Tim Parenti replied to Sorel Si:

> The zone in question was moved from the 'europe' file to 'backzone' in
> release 2022d and, as such, is no longer referenced in the 'zone*.tab'
> files.
> ...
> Any remaining references to Europe/Uzhgorod exist for backwards-
> compatibility purposes only.  In almost all cases, Europe/Kyiv should
> be used instead.

There’s another problem, though. The tz database is not a gazetteer or other 
official reference to the “correct” spelling of city names. The identifiers in 
the tz database are not meant to be presented directly to users as part of a UI.

Rather, tzids are internal identifiers, sort of like variable names or URLs or 
API endpoints. They are intended to be stable. When you change one, you create 
an incompatibility with systems that expect the old name.

The theory file says to “Use mainstream English spelling” in place names, which 
often leads to requests like this on the basis that the locally preferred 
spelling or transliteration must automatically be the mainstream English 
spelling. That’s not reality. It takes time for another government’s preference 
to take hold in English usage—we still don’t really use “Czechia” in the US, 
and might never do so—but even if and when it finally does, the stability 
problem remains.

Although the full tz database includes a system of links, such that clients 
looking for an old tzid can have it automagically mapped to the new one, there 
are systems out there that don’t use this mechanism. They look up the names 
directly, or they provide mappings between tzids and time zone names used by 
other systems. (CLDR is an example of this.) When a tzid is changed, it breaks 
the client, at least until the maintainers of the client get a chance to react 
with an update.

The theory file goes on to say, “If a name is changed, put its old spelling in 
the 'backward' file as a link to the new spelling. This means old spellings 
will continue to work.” This is great for clients that understand ‘backward’ 
and linking, but breaks the others. I would prefer to see the administrative 
policy changed in the opposite direction that Sorel has in mind, making the 
_new_ spellings the aliases, and keeping the existing identifiers stable for 
systems that need that stability.

(Lest there be any misunderstanding, this has nothing to do with being 
pro-Ukraine or pro-Russia.)

--
Doug Ewell, CC, ALB | Lakewood, CO, US | ewellic.org

Reply via email to