From twinsun!twinsun.com!eggert@CS.UCLA.EDU Tue Oct 19 18:07:51 1993 Return-Path: <twinsun!twinsun.com!eggert@CS.UCLA.EDU> ... From: eggert@twinsun.com (Paul Eggert) Message-Id: <9310192204.AA24829@spot.twinsun.com> Date: 19 Oct 1993 15:04:24 -0700 To: ado@elsie.nci.nih.gov In-Reply-To: <9310192039.AA26353@elsie.nci.nih.gov> (ado@elsie.nci.nih.gov) Subject: Re: Shanks's International Atlas ...
That being so, we'd want to be on the lookout for rare timekeeping changes and be sure to include them.
Yes, definitely. (Ireland was an eye-opener for me.)
2. A pragmatic answer to how many zones to have for France is "as many as are needed to describe the present day zones." As you noted, though, doing even this at the country level yields lots of zones.
If we were to do this, we'd need a convention to help decide which of the 153 zone histories we'd present as `France'. I see two plausible options: pick the present-day capital (i.e. Paris), or pick the most typical and widely-used (for France, this would probably be Shanks's French Time Zone #1, which I believe is the time zone history of the un-fought-over boondocks during the two world wars).
3. As a last resort, simply reporting UCT for a zone when it's doing something awkward is a possibility.
If by `awkward' you mean LMT, I'd prefer using standard time instead, since it's less of an error. E.g. if we use the district's present-day capital for (2), then we could use the capital's LMT for times before standard time was instituted.
If we use present-day capitals, I think it'd be far more honest to give time zone files names like `Paris' and `Berlin' instead of `France' and `Germany'. This is reminiscent of the interface used in the HP 100LX palmtops -- they ask you what city you're in, not what country you're in.