Yes, it would be nice to use the binaries, but I have to provide a complete solution that can, upon someone retrieving and extracting the tz data, update the db with a click of a button on a Windows computer. So I have to compile the data myself. If the format of the data files changes, that's fine...I'll just update the process. It doesn't seem too difficult, though. Reading some of the messages in the archive has helped me to better understand the process. I basically have to create UTC transition dates for every year covered by an individual rule. I create a date/time from the rule, adjust it by the GMTOFF of the zone (accounting for u indicators,) and then adjust it by any daylight saving time that might be in effect from a previous rule (accounting for s indicators.) That's not so bad. I'm still trying to figure out a slick way of handling max, though. On Mon Oct 24 05:23:23 2011"Alois Treindl" <alois@astro.ch> wrote:

Personally, I would not recommend of writing yet another tz compiler.

Reading and processing the binary files created by zic seems to me a much cleaner and more reliable approach.