[time-nuts] [Solved] Looking for multiple PPS timestamp logging

Kevin Rosenberg kevin at rosenberg.net
Wed Oct 5 20:05:25 UTC 2011


Hi Chris,

select is a great idea, but I think I'll be using cygwin which doesn't
appear to have an emulation of select on Windows.

Kevin

On Oct 4, 2011, at 10:23 AM, Chris Albertson wrote:

> On Tue, Oct 4, 2011 at 3:36 AM, Hal Murray <hmurray at megapathdsl.net> wrote:
>>> The temperature swings won't be large, just the usual diuneral indoor cycles.
>> 
>> In California, the diurnal temperature swings are big enough to be useful. :)
>> 
>> This is what ntp sees the main/CPU crystal doing in a non airconditioned
>> room.  Each color is a different day.
>>  http://www.megapathdsl.net/~hmurray/ntp/slope.gif
>> 
>> The swing would be much bigger outside, especially with an open sky.
>> 
>> 
>>> Since we'd be measuring 4 PPS signals that won't be in phase, I wasn't
>>> planning on sleeping for long. But, I some 10 usec sleeps with usleep(10000)
>>> after reading a pulse would be reasonable. But, with the system not doing
>>> anything  but polling and writing to the disk, I wasn't expecting much
>>> interruptions from other processes.
>> 
>> I was thinking of keeping track of when you expected each pulse to arrive
>> next and sleeping until a little before you expected the soonest one.
>> (Adjust "a little" by trial and error.)
> 
> This is a UNIX-like system?  If so use a "select" on the file
> descriptor(s).    This system call allows you to sleep until there is
> data available on one of a set of files.   See "man 2 select"
> 
> The code right after the select() has to poll all the files to see
> which one has data, your read/process it then go back an wait/sleep.
> If you do poll after a fixed length sleep you will on average to 1/2
> of a polling cycle late reading the data and also you waste time
> checking when nothing is there.
> 
> Chris Albertson
> Redondo Beach, California
> 
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
> 




More information about the time-nuts mailing list