I wrote:
I was thinking about adding a function to our code that called clock_gettime() (which is what our old code did), but then constructing a time_type out of that using the UNIX epoch for the date part and the results of clock_gettime() for the time_duration part, i.e, in pseudo code:
time_type our_get_time() { clock_gettime(CLOCK_REALTIME, ×pec); time_duration td = massage(timespec); return time_type(unix_epoch, td); }
I ended up doing something like this: struct timespec ts; clock_gettime(CLOCK_REALTIME, &ts); time_duration td(seconds(ts.tv_sec) + nanoseconds(ts.tv_nsec)); return ptime(epochDate, td); where: const date epochDate(1970, Jan, 1); Wow...this turns out to be much, much slower, like 1000x slower, than calling gettimeofday() followed by gmtime()......! I guess it must be all the calculations involved to normalize the very large seconds value offset from 1970 into a ptime. Hmmm...back to the drawing board.