Paul Eggert wrote:
Lester Caine wrote:
I'd almost argue that NO historic data is better than a SINGLE subset arbitrarily selected? Currently there is no mechanism to tell you that the historic data may not actually apply at all?
These are both valid points. It's certainly easy to remove data before 1970 automatically, and some platforms do that now (QNX, for example), so the first option is already there and some platforms already take it.
As for the second, we could add a comment to each zone specifying the earliest time stamp such that the zone is known to be valid for all locations in its region after that time. The default would be 1970, which means the tz database trivially conforms to this pattern now.
I see two problems with this idea, though. First, it needs to be integrated with winnowing, and it's not yet clear how that would work. Second, and more important, except for Europe/London and perhaps a handful of other places, we wouldn't know what value to put into the comment, other than the default "1970". This seems a fairly serious objection.
I think we have reached the point where it's clear that TZ is never going to support anything before 1970 with any reliability. winnowing is a pointless exercise unless you are prepared to add new data later. So the only SAFE way to distribute a sub-set of TZ data is not to link it to past history that it is not 100% related to. The only problem remaining is how to allow a correct historic record to be used with the restricted TZ data. -- Lester Caine - G8HFL ----------------------------- Contact - http://lsces.co.uk/wiki/?page=contact L.S.Caine Electronic Services - http://lsces.co.uk EnquirySolve - http://enquirysolve.com/ Model Engineers Digital Workshop - http://medw.co.uk Rainbow Digital Media - http://rainbowdigitalmedia.co.uk