[tz] Scheduling - use case for acronyms

[random832 at fastmail.us]

On Tue, Apr 2, 2013, at 15:11, Philip Newton wrote:
> On 2 April 2013 19:42, Tobias Conradi <tobias.conradi at gmail.com> wrote:
> > Exactly, and that would be communicated how, in a brief manner - as
> > Mark used in "Mondays at 8:00 PT"
> 
> On the other hand, if the meeting took place in Berlin, some would say
> "Mondays at 8:00 CET" while others would say "an Montagen um 8:00
> MEZ".
> 
> Which is a case for placing such time zone abbreviations in
> localisation frameworks such as CLDR rather than in the tz database.

Yes, it is. And, in fact, there are time zone abbreviations in CLDR.

However, there are no C implementations that natively support the time
zone abbreviations in CLDR - in no small part, I suspect, due to the
lack of any support for either localized timezone abbreviations or
metazone specification in the tzfile binary format.

Metazones, for those who don't already know, are things that group zones
that are "the same timezone" (e.g. the two dozen or so U.S. and Canada
-05:00 EST/-04:00 EDT zones are all "America_Eastern", your
CET/MEZ/whatever are all "Europe_Central", etc) together, and are the
key used to look up the localized timezone names and abbreviations. They
include a standard and daylight offset, and of course a name, but _not_
the question of whether DST is used at all, or what the specific
changeover dates for it are.

The data is at
http://unicode.org/repos/cldr/trunk/common/supplemental/metaZones.xml,
but note that it does not cover changes before 1970 (for example,
America/Indianapolis is simply mapped to America_Eastern, rather than
showing it should be America_Central before 1955 Apr 24 and from 1957
Sep 29 to 1958 Apr 27.)

Really, metazone specification itself seems like it _should_ be the
domain of tz rather than CLDR. But it would require an incompatible
extension to the syntax of the timezone source files.