Re: [gpsd-dev] Detecting Raspberry Pi Hat & PPS

[Gary E. Miller]

Yo Frank!

On Fri, 6 May 2016 09:42:58 -0400
Frank <address@hidden> wrote:

> I just did a fresh git clone (8:30AM EST).  Recompiled & the first
> run, gpsd sees PPS:

Weird, the only change was to a log message!  That would imply a wild
pointer or some such...

> gpsd -nND 9 /dev/ttyAMA0 |& fgrep PPS

[...]

> gpsd:PROG: KPPS:/dev/pps0 assert  1462537682.901482605, sequence: 53362, 
> clear   0.000000000, sequence: 0 - using: assert

Yeah, good.

> However… 
> It consistently works when starting with:  gpsd -nND 9 /dev/ttyAMA0
> I’ve narrowed it down to if it’s started *without* “-n”, then there’s
> never any attempt to even look at PPS (no PPS debug entries) &
> cgps/gpsmon do not see PPS

You ALWAYS need the -n.  I have complained to Eric about this for years.

gpsd does nothing without -n, until a client connects, but with a time
server you do not expect to see a client.  This is by (flawed) design.

I'll not go into the bad data that led to this bad decision.

> > Checking for C header file sys/timepps.h… yes  
> 
> Note that the troubleshooting before today (May 6th), gpsd was built
> before the “timepps.h” file was on my system.

Well, that would be your problem right there.  So not a wild pointer,
but an insufficient build.  The new log message will point that out.

I suspect scons should alos print a loud warning.

RGDS
GARY
---------------------------------------------------------------------------
Gary E. Miller Rellim 109 NW Wilmington Ave., Suite E, Bend, OR 97703
        address@hidden  Tel:+1 541 382 8588

pgpCRZHWBT0xp.pgp
Description: OpenPGP digital signature