Jan. 15, 2020
8:24 a.m.
On 1/14/20 7:04 PM, Steve Allen wrote:
the original draft of CCIR Recommendation 460 allowed for leaps of "integral multiples" of one second.
Ah, I didn't know that. Still, the 1989 C standard botched this, since C89 allowed for leaps of one or two (but no more) seconds, apparently on the mistaken assumption that that could have happened under the then-current leap second rules. And this botch was copied into POSIX. Although the two-leap-second botch was fixed in the C99 and POSIX-2001 standards, it was still present (though with bugs) in tzcode until the buggy support was removed in tzcode 2017c.