Re: [gpsd-dev] errors from scons testregress: Trimble-lassen

[Eric S. Raymond]

Hal Murray <address@hidden>:
> I'm not familiar with the gpsd build/test environment.  On a recent git pull, 
> scons (build) seemed happy, but scons testregress gives the following errors:
> 
> Regression test FAILED: 3 errors in 78 tests total (0 not found).
> The following test Failed:
> ================================================================
> "test/daemon/trimble-lassen_iq-3dfix.log"
> "test/daemon/trimble-lassen_iq-playacar.log"
> "test/daemon/trimble-lassen_iq.log"
> ================================================================
> Elapsed time: 281
> scons: *** [gps-regress] Error 3
> scons: building terminated because of errors.
> 
> 
> Looking way back, I see something like this for all 3 cases:
> 
> Processing test/daemon/trimble-lassen_iq-3dfix.log
> Traceback (most recent call last):
>   File "/home/murray/gpsd/gpsd-git/gpsfake", line 171, in <module>
>     test.gps_add(logfile, speed=speed, pred=fakehook)
>   File "/home/murray/gpsd/gpsd-git/gps/fake.py", line 453, in gps_add
>     progress=self.progress)
>   File "/home/murray/gpsd/gpsd-git/gps/fake.py", line 249, in __init__
>     [iflag, oflag, cflag, lflag, ispeed, ospeed, cc])
> termios.error: (22, 'Invalid argument')
> --- test/daemon/trimble-lassen_iq-3dfix.log.chk 2011-10-24 01:29:33.000000000 
> -0700
> +++ /tmp/gpsd-test-uK66kjQzdvfWn0/test-4743.chk 2012-05-05 18:24:53.000000000 
> -0700
> @@ -1,251 +0,0 @@
> -$GPGSV,3,1,10,00,00,000,00,19,26,250,29,00,00,000,00,18,26,096,37*7C^M
> -$GPGSV,3,2,10,01,29,221,26,11,24,307,29,03,08,225,28,00,00,000,00*7B^M
> -$GPGSV,3,3,10,22,65,094,37,09,24,045,37*78^M
> -{"class":"SKY","tag":"ID5c","xdop":1.21,"ydop":5.27,"vdop":2.56,"tdop":1.47,"
> hdop":1.43,"gdop":3.28,"pdop":2.93,"satellites":[{"PRN":19,"el":26,"az":250,"s
> s":29,"used":true},{"PRN":18,"el":26,"az":96,"ss":37,"used":true},{"PRN":1,"el
> ":29,"az":221,"ss":26,"used":false},{"PRN":11,"el":24,"az":307,"ss":29,"used":
> true},{"PRN":3,"el":8,"az":225,"ss":28,"used":false}]}^M
> ...
> 
> I'm running on a old Fedora box.  Kernel is 2.6.38.5 with my local config 
> file.
> 
> Is this a known problem area?  If not, what do I do to help track it down?

This is a bug in older versions of the Python termios library, I think.  
What's triggering it is gpsfake trying to set a word-length/stop-bits
combination other than N8, which those device types and only those require.

I dimly recally tripping over something similar in the past.  Simplest way
to fix it is just upgrade your Python version.
-- 
                <a href="http://www.catb.org/~esr/";>Eric S. Raymond</a>