The apparent-solar-time zones seem to be misusing the "zzz" indicator, and they have poor boundary behaviour. The attached patch makes these changes that seem like a good idea to me: * abbreviation "LAT", for "Local Apparent Time", during the period where the zone provides LAT behaviour * additional apparent-time offset entry for Dec 31 of previous year, so that the ruleset provides LAT correctly for the first half of Jan 1 * range of validity of the LAT data defined strictly as starting and ending at Jan 1 00:00 UT * zone marked as disused outside the range of validity of the LAT data Also, thoughts on apparent solar time more generally: the existing solar?? files inconveniently can't be merged because the result would have more than 256 distinct offsets. However, if the resolution of the offsets were dropped from 5 s to 10 s, then there would be only 186 distinct offsets within the range of the equation of time, so we could have an apparent-time ruleset covering as many years as we're interested in. That could allow incorporating those rules into the regular Asia/Riyadh zone, rather than requiring these hacky separate zones. The apparent-time ruleset could also be formulaically generated at build time, rather than explicitly included in the source tarball. -zefram