Ton Voon wrote:
>server's timezone to CET.

Dodgy concept there.  It really doesn't make sense for a server as a
whole to have a default timezone other than UT.  A timezone choice is
a feature of certain types of application data, not a feature of where
you run the application.

>I can understand the arguments about EST, but if there's no ambiguity  
>about CET,

We just found an ambiguity about CET: does it include the DST rules?
Also, consider that your form of CET, with European-rules DST, is only
defined for 1977 and later (when the EU DST rules are defined), so it
falls apart if you need to process earlier times.  For those applications
you're forced to distinguish between Europe/Paris, Europe/Berlin, and
the others.  Not to mention that countries went onto CET-with-DST at
different times.

If you have some definite idea about what "CET" means in your
application context, go ahead and implement it.  That might be an alias
to Europe/Berlin, or a fixed offset from UT, or a synthetic timezone
(with its own tzfile) that doesn't match civil time anywhere.  But be
aware that it's not universally meaningful, and do it at the application
level without gumming up the implementation for everyone else.

>                                               I note that EST, MST  
>and HST are supported timezones: 

Historical oddities.  Not role models.

-zefram

Reply via email to