Thank you all. This is what I understood as well. On 2013-07-17 13:32, Tim Parenti wrote:

On 17 July 2013 12:53, David Patte ₯ <dpatte@relativedata.com <mailto:dpatte@relativedata.com>> wrote:

Am I correct to say, that for every lat/lng there (theoretically) exists a zone in that country and/or region that represents its timezone rules since 1970; that if no tz zone exists in that country and/or region that represents it clock rules since 1970 correctly, that a new zone is to be created in tz for it?

In theory, yes, that is my understanding. Of course, in international waters, the zone that "correctly" represents the local rules would be, e.g., Etc/GMT-3, but in general, if an existing set of rules does not properly cover a region (for all timepoints since 1970), a new zone is to be created.

On 17 July 2013 12:53, David Patte ₯ <dpatte@relativedata.com <mailto:dpatte@relativedata.com>> wrote:

I am looking at this from the perspective of timezone boundaries; that tz boundaries (theoretically) don't change unless new tz zones are created, or it is discovered that a latlng's clock 'rules' since 1970 can be more accurately be described by a different tz zone?

The tz project neither defines nor tracks zone boundaries; however, it is sometimes helpful to think of zones in this manner. Obviously such boundaries would change upon the creation of a new zone. You could make the case that such boundaries would also change if it is discovered at some point that location A better matches zone X than zone Y, but as long as location A is not itself defined in zone.tab, this does not affect tz data, and thus does not concern the tz project beyond documenting appropriate commentary.

-- Tim Parenti

--