12:03 < hoyhoy> I'm going to point a web cam at my wall clock and use pymorph to sync it to an ntp server with the help 
of a
                stepper motor
12:03 < hoyhoy> who's with me?
12:03 <@gregh> I think you should do it the other way and create an ntp reference source based on your wall clock
12:03 < hoyhoy> that wouldn't be challenging enough
12:04 <@gregh> bet you could make slashdot with it though
12:04 < hoyhoy> I think the other way would be more of a slashdot thing
12:05 < hoyhoy> another consequence of that is that your wall clock automagically adjusts to daylight savings time.
12:05 <@gregh> now that would be spiffy
12:05 < hoyhoy> it has to be invisible though
12:06 <@gregh> no wires, hook it up with bluetooth
12:06 < hoyhoy> what's the smallest stepper motor available?
12:06 < hoyhoy> that'd be some hellacious PIC microcontroller work though.
12:06 <@gregh> heh
12:07 <@gregh> so do the software on a pc and just send step signals to the clock like an RC car
12:07 < hoyhoy> ooooh, I have some zipzap cars around.
12:08 < hoyhoy> there's an RPM to hook the controls into a lunix box
12:08 < hoyhoy> and the reciever is ridiculously small
12:08 < hoyhoy> probably would still need a PIC though
12:09 < hoyhoy> to interface to the stepper motor
12:11 < hoyhoy> I doubt the zipzap motor is strong enough to turn the clock even if it was possible to sync up a certain
                duration of engine rotation to a certain amount of time.
12:11 < hoyhoy> I suppose I could make some kind of gear mechanism to turn the clock ridiculously slow.
12:11 < hoyhoy> like 10 seconds of rotation = +1 minute
12:12 < hoyhoy> but the 1/3AAA battery would likely run out.
12:14 <@TheCougar> actually, you can change the oscilator that drives the clock
12:14 <@TheCougar> model railroaders do it to simulate a full day in a few hours
12:14 <@TheCougar> you could use a VCO that could be commanded via bluetooth
12:15 <@TheCougar> as it starts to fall out of sync, you speed up or slow down the oscilator a bit
12:16 < hoyhoy> I think it would be better to do an update every 6 or 12 hours
12:16 < hoyhoy> I don't want to have to run AC to the clock
12:17 < hoyhoy> every 6 hours sync the wall clock time.nist.gov
12:18 <@TheCougar> who's talking about running AC to the clock?
12:18 <@TheCougar> and btw, any clock that uses AC as a timebase never needs to be reset (other than for DST)
12:18 < hoyhoy> yeah
12:19 <@gregh> in theory :)
12:19 < hoyhoy> well, constant updates over RF would probably take it's toll on the battery.
12:19 < hoyhoy> I guess the clock is only rxing
12:20 < hoyhoy> still, power is a concern
12:20 < hoyhoy> I think a 6 hour rule or on a CDT/CST boundary would be sufficient.
12:20 < hoyhoy> for syncs
12:22 < hoyhoy> maybe just daily updates at 3 AM.
12:22 < hoyhoy> yeah, then that would CDT<->CST
12:23 <@TheCougar> depends on how accurate you want to be and how badly the clock drifts
12:24 < hoyhoy> then I can create charts that calculate clock drift per day, and have the sync algorithm and determine 
the
                differential equations that equate clock drift to battery drain.
12:25  * Nugget watches hoyhoy reinvent xntpd
12:25 < hoyhoy> then absolve the need for pymorph after a certain amount of data collection.
12:26 < hoyhoy> then I can go into business selling analog clocks that sync to ntpd