GPS as a source of time like NTP
sears at cs.berkeley.edu
Sun Jul 27 08:29:29 CEST 2008
Daniel Willmann wrote:
> Just had a look at the gpsd sources. There's a gpsd/refclock_nmea.c in
> there. I guess it should be pretty easy (at least doable :-) to make a
> refclock_dbus/gypsy.c and listen to the specific dbus signals.
Getting this to work would be great! My FR drifts by a few minutes
every day. I think ntp has code to adjust for this so the clock won't
jump around every time it talks to a high-quality time source.
Also, I've heard that GPS satellites don't take leap seconds into
account. That could cause trouble for people using ntp and gps data to
set clocks, since the times will disagree by a few seconds. It might be
good to map from "GPS time" to "real time".
More information about the devel