[gpsd-dev] Please test for a specific bug:

[Ed W]

I don't have the ability to run tests on GPSd pre-release (at least 
likely before you release it).  However, most of the previous versions, 
including 3.10 show a problem detecting very low rate AIS input sentences

Specifically:
- Given an AIS in a weak area (we can *just* see the Thames river from 
our location). Occasionally many minutes can go by without seeing a 
sentence.
- I have a pre-detection service triggered by udev which scans any new 
serial ports and tests a few baud rates to see if they emit NMEA
- My script sees NMEA at baud 38400, and then passes the device to gpsd 
(so definitely there were AIS sentences on this input and likely we left 
the port configured to 38400 baud)
- *Unless* there is another AIS sentence emitted fairly quickly gpsd 
seems to start to skip testing for AIS/38400 input forever more?  it 
then seems never to find that there is NMEA on this device and appears 
to never detect anything (I realise this is a bit woolly, I don't have a 
test system at hand to be more precise)
- Note this testing is complicated by the infrequency of NMEA sentences, 
ie it's theoretically possible that if I just left it for "more hours" 
it might eventually detect the AIS, but my point is that "10s of minutes 
go by" and the AIS isn't picked up, despite that sentences are being 
emitted.


I believe a test case would be as simple as:
- Connect a serial to usb adaptor (with nothing connected to it)
- Add the serial port to gpsd
- Wait some minutes (say 5 minutes)
- Now connect the AIS to the serial port

If this works then I'm happy that this is resolved.  Although note that 
there might simply be a "slow detection" issue here if the AIS sentences 
come at a sufficiently low rate, ie gpsd is simply latching onto the 
garbage that comes when the baud rate is wrong..

I realise this is a poor report of a potential problem, but it's the 
best I can do from my current position... If it helps keep that defect 
count low though!

Kind regards

Ed W