[LEAPSECS] Time math libraries, UTC to TAI

Zefram zefram at fysh.org
Sat Jan 7 12:51:20 EST 2017


Steve Summit wrote:
>great for printing human-readable nanoseconds, but if you want to
>print milliseconds or microseconds, you have to extract it and
>manipulate it yourself.

$ date +"%3N %6N %N"
198 198364 198364083

It's an irregular treatment of field width, but admirably satisfies the
obvious needs.

The only other mode of use for tm_nsec in strftime that I can see would
be sexegesimal division of the second.  Nanosecond resolution is slightly
finer than fifths, so that's three more format specifiers required if
you care to implement them.

>Yeah, especially since the promised zettabyte drives are still
>nowhere in sight, and we're all having to make do with these
>cramped old terabyte ones.

Well, quite.  We *are* using a lot of this newfound space to store
unprecedented numbers of files, so the four bytes per inode multiplies up.
But the rise in numbers of files is not in proportion to the increase
in capacity.  The whole exercise does seem to be optimising for the
wrong thing.

-zefram


More information about the LEAPSECS mailing list