On Mon, 13 Jan 2003, John Peacock wrote:

> This is a very good point, and another reason to consider TAI64 as the base
> storage.  One thing to be careful about is to not fall into the trap that
> localtime() and xntpd have with regard to leap second handling:
>
>       http://cr.yp.to/proto/utctai.html
>
> If we want to commit to subsecond times, we would have to use TAI64NA, but we
> could hide that from the general implementation by only exposing the additional
> precision when requested.

I'll commit to libtai64 when it runs on all the platforms Perl does.  Til
then, I want to have a reference implementation in Perl.  An
implementation that uses libta64 is still very welcome as an faster/more
precise (I assume) alternative for those who can take advantage of it.


-dave

/*=======================
House Absolute Consulting
www.houseabsolute.com
=======================*/

Reply via email to