Re: gpsprof -r fails with no gpsd running

[Greg Troxel]

"Gary E. Miller" <address@hidden> writes:

> Yo Greg!
>
> On Thu, 23 Jan 2020 10:45:06 -0500
> Greg Troxel <address@hidden> wrote:
>
>> (I'm trying to fix the gpsprof problem with TPV records with no fix.)
>
> Cool.  Another nice to have feature would be an option for how many minutes
> to capture.

Could be, but I always want the file and don't really trust gpsprof at
this point, so I use gpspipe -w to capture and then reprocess until I
have debugged gpsprof.

>> Running gpsprof with a saved file results in a stack backtrace in the
>> case where the is no runnning gpsd.
>
> Yes, I do that.

As in you tried this and got the same error, with 3.20 or master?

Or ?

Trying to make sure it fails for you too before bisecting, but I bet it does.

>> The problem seems to be that it creates gps.gps(None) but that tries
>> to connect anyway.
>
> This part of gpsprof gets broken every few years...
>
>> $ python3.7 gpsprof -r  <  saved-logfile-gpspipe-w-style.txt
>
> Why do you specify python3.7?

Because the python path passed to the build didn't get substituted into
the gpsprof script in my git clone, and I was trying to run that vs the
installed 3.20, and was doing too many things at once.  Not important.

>> gpsprof: replotting...Traceback (most recent call last):
>>   File "gpsprof", line 1179, in <module>
>>     plot.replot(sys.stdin)
>>   File "gpsprof", line 321, in replot
>>     self.session = gps.gps(host=None)
>>   File "/home/n0/gdt/SOFTWARE/GPSD/gpsd/gps/gps.py", line 231, in
>> __init__ gpscommon.__init__(self, host, port, verbose, reconnect)
>>   File "/home/n0/gdt/SOFTWARE/GPSD/gpsd/gps/client.py", line 43, in
>> __init__ self.connect(self.host, self.port)
>>   File "/home/n0/gdt/SOFTWARE/GPSD/gpsd/gps/client.py", line 69, in
>> connect self.sock.connect(sa)
>> ConnectionRefusedError: [Errno 61] Connection refused
>
> Oh, that is new-ish.  Maybe related to the reconnect patches?
>
> Maybe try a bi-sect?

OK.  I will first try to figure out if broken in 3.20, and then bisect.

>> I am unclear on whether gps.gps(host=None) (see line 320, or close,
>> within replot()) is
>> 
>>   wrongly called because it assumes a running gpsd, or
>> 
>>   wrongly behaving, in that None should produce a valid object that
>> does not try to connect and always behaves as if disconnected, or
>> 
>>   something else
>
> My guess is the meaning of None got changed somewhat recently.

As in gps.gps() use to take host=None to mean "do not try to connect",
and now it means localhost somehow, or something like that?