[Sbcl-devel] get-internal-real-time and leap seconds

[Ilya P. <ipe...@dw...>]

Hi all,

I got bitten by the last leap second occurrence (on Dec 31). The Linux
kernel stepped system time by a second, my code measured a negative
time interval using get-internal-real-time, and died.
get-internal-real-time uses "gettimeofday", which is affected by jumps in the system time.
Do it make sense to utilize "clock_gettime (CLOCK_MONOTONIC, ...)" in
get-internal-real-time (on OSes where CLOCK_MONOTONIC is available)?

Thanks,
Ilya