Date: Mon, 25 Jun 2018 10:16:05 +0100 From: Lester Caine <lester@lsces.co.uk> Message-ID: <cf68c796-2db1-a653-262b-3a55aeeb27f7@lsces.co.uk> | The whole system SHOULD return 'don't know' where data is simply missing | for whatever reason, You can design a whole set of new interfaces with that kind of return available if you like. Traditionally localtime() was expected to always succeed (there was no error return possible at all) - that needed to be altered when the "int" tm_year became too small to translate all time_t's but that's the only error now. | and using current data blindly with historic | timestamps will always be wrong! Using current data for future timestamps is just as likely to be wrong. But regardless of that, getting something relatively close to correct is adequate for 99% of the uses that exist. For historic values, while the zone data might not be correct, the recorded timestamp is just as likely to be even more incorrect (or vague) so the loss there is not as great in most cases as it seems it could be. kre