The CRC check doesn't catch all bad messages

[lesamouraipourpre]

Because, the CRC value in the NMEA sentence is only 8-bit, it is fairly easy for a bad message to get past the CRC check.
By observation, with my GPS on the window it takes on average ~3 hours before a bad message crashes the program.

This is usually triggered by _parse_int and _parse_float functions and can happen elsewhere.

[lesamouraipourpre]

I intend to fix this by making the _parse_int etc. methods always return either a valid value or None and never raise a ValueError.

I also intend to put checks into the _parse_gpgll etc functions which will fail quickly if the wrong number of comma separated parts has been received.

[tannewt]

Would it be possible to fail in the same way as CRC if parse int or float fail? That way you still get an error but you only have one type of error to handle.

[lesamouraipourpre]

I'm aiming to keep it so that the update() method behaviour stays the same, specifically
Returns True if new data was processed, and False if nothing new was received.
Returning False in the event of an unparseable sentence will match the behaviour of a CRC failure.

[tannewt]

Ok, that sounds perfect! Thank you!