Re: [gpsd-dev] the plot thickens on OS X

[Frank Nicholas]

On Sun, Feb 15, 2015 at 7:45 PM, Greg Troxel <address@hidden> wrote:

I am livetesting with a GR601-W (ublox 6), bisecting.   I find that the
first open of the device after it has been plugged in behaves
differently from subsequent opens.  Subsequent opens almost always get
the "no probe matched" that I sent before.   First opens sometimes work,
and sometimes loop with lots of reads of a few bytes.

The way to get it to work seems to be:

  plug in USB
  wait about 2-6 seconds
  start gpsd

starting it just about when the blink starts can lead to log spewing
with it getting full buffers and not recognizing anything.

I am unable to make it work reliably by any specific timing.

This doesn't seem super different before/after that "restore blocking"
commit you pointed me to.

Something really looks amiss with the ublox lockin code, from reading
the logs.   There are lots of "unknown char" with eg 647 chars in the
buffer, looking like beaddeef which is suspicious.  But I also see what
looks like nmea at times, but still it doesn't lock.

This sounds similar to what I was seeing with my Bluetooth Sirf III GPS.  It was extremely inconsistent, and unless I power cycled the GPS, gpsd would not sync at all.  I attributed it to bluetooth->Serial issues.  Maybe I was wrong...