[Zope3-dev] Strawman Proposal: Timezone as a location

[Guido van Rossum]

> I wasn't trying to and I do believe you. I guess I just thought the
> notion of 25 location-based timezone values was intrinsic enough to
> include in the basic definition of things. But now I realize that
> this would be easy enough to implement in a subclass of tzinfo,
> therefore there is no need for what I propose.

(1) Not all tzoffsets are a whole number of hours from UTC
    (e.g. Newfoundland which was mentioned here a moment ago is
    apparently 3.5 hours behind UTC).

(2) Your use of "timezone" is still ambiguous.  If you use it to mean
    tzoffset (i.e. a number of seconds east of UTC), that definition
    doesn't support any kind of daylight saving time, and that makes
    time calculations unintuitive when they cross a DST jump.  If you
    use it as the name for a particular jurisdiction's set of rules
    for calculating its tzoffset for any date/time, there are way more
    than 25 timezones (e.g. the southern hemisphere has DST from
    October through some time in February-April).

> Is that what you're trying to get me to see? (Go ahead and hit me on
> the head with a sledgehammer if necessary. I'm worn out and need to
> be bludgeoned with the obvious.)

Go to bed.  It's 02:19 UTC. :-)

--Guido van Rossum (home page: http://www.python.org/~guido/)