Karel Zak <kzak@xxxxxxxxxx> writes: > Applied with a small change, I have renamed --predict-hc to --predict > (because we have --set, --show, etc. commands). Thanks. > I see a small nit here. You can specify --date from past: > > # hwclock --predict --date "1/1/2000 16:00:00" > Sat 01 Jan 2000 03:59:49 PM CET 0.376034 seconds > > not sure what this prediction means :-) Maybe we need to check that set_time > value is greater than gettimeofday(). I'd keep it without that check. If I log an RTC reading every midnight at a GPS time pulse it is useful to be able to plot how well the linear prediction used by hwclock matches historical data. Some background: http://lindi.iki.fi/lindi/openmoko/compare-clock-sources.pl is a script that I used on openmoko to compare different time sources. On my openmoko RTC drifts almost 5 seconds every hour. For this reason I realized it is simpler if I do not write to RTC at all. I let it float freely and just have function that maps real time to this imaginary RTC time. Since I never write to RTC I can take an RTC reading every midnight (by GPS time pulse) and save these to a file to later monitor how well the linear prediction works (so far it has been great!). If I constantly wrote to RTC I do not see how I could easily do such long-term monitoring. -- To unsubscribe from this list: send the line "unsubscribe util-linux-ng" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
