[LEAPSECS] USWP7A docs for 2013 September meetings

Zefram zefram at fysh.org
Tue Aug 20 06:56:47 EDT 2013

Warner Losh wrote:

>By catering to the sucky implementation, it precludes the precision one :(

Doesn't preclude it. The only real omission around the Clock class is
that the API provides no way for a Clock to advertise its accuracy,
precision, stability, or related features; accuracy is only claimed
out-of-band. So you could say that the API precludes *recognition of* the
precision implementation. I suspect that this will be the first thing to
be rectified in the next round of standardisation, following real-world
experience with JSR-310. (You need a method on Clock that yields the
current Instant plus an inaccuracy bound applying to that reading, not
(only) a separate method that yields a non-specific accuracy parameter.)

Likewise, precision computation with points represented on the Java
time scale isn't precluded, it's just out of scope for the standardised


More information about the LEAPSECS mailing list