Re: [gpsd-users] Earthmate and the Zodiac Driver

[Eric S. Raymond]

Scott Nolde <address@hidden>:
> gpsd:PROG: selecting Generic NMEA driver...
> gpsd:PROG: => Probing for FV-18
> gpsd:WARN: unknown sentence: "EARTHA"
> gpsd:PROG: found trigger string EARTHA.
> gpsd:PROG: switch_driver(Delorme EarthMate (pre-2003, Zodiac chipset)) 
> called...
> The Earthmate reliably worked before the 3.x change.

Yeah.  It's switching drivers all right, so the recognizer is working...
but it's not staying locked on. I can't tell why from the output.

Unfortunately, the old Earthmate is one of a handful of devices we
support that cannot be simulated with a canned test load (that's due
to the EARTHA handshake). 

If it were simulable, the change that broke it would not have gotten
past our regression tests.  As it is, the only way to isolate the
fault is to test with live hardware.  This means you'll have to do a
bisection.

A bisection works like this: 

1. You clone the git repo.

2. You pull 2.96 and verify that it works.

3. You check out a revision halfway between the known-good revision and
   the first known-bad one.

4. Test.  If it works, this revision becomes the last known-good one.
   If it doesn't, it becomes the first known-bad one.

5. Return to step 3.

Bisection terminates when you get differing results for two adjacent
commits.

You should be able to isolate the bad change in about 12 tests.  If you
can do that, I can fix it.
-- 
                <a href="http://www.catb.org/~esr/";>Eric S. Raymond</a>