[gpsd-dev] Much was accomplished today at Asylum House

[Eric S. Raymond]

Much was accomplished today at Asylum House. In about five hours of
work, interrupted by breaks for excellent Indian takeout and
Eurogaming with our lovely and remarkably tolerant wives, we advanced
towards several NTPsec goals and learned at least one thing of
probable interest to GPSD developers.

The three test-farm Raspberry Pis all got properly cased, an adventure
which involved my first-ever use of a Dremel tool (to make apertures
for the protruding SMA jacks). The reason I judged this to be
necessary happens to be padding around on my desk as I type. Voltages
a human could shrug off could be dangerous to critters with the
smaller body mass of cats, and cat hair is not good for electronics.

All three also now have internal standoffs from Phil's parts bin, the
the better to withstand drop shocks.  It is, alas, not that uncommon for
them to fall the two feet to my desk from the Official Windowsill of
Mad Science.

Image: 
https://photos.google.com/photo/AF1QipO-7c9GmGY-US86pA6RxdYLcOymT1gNLqqYgBCj

I got the HAT for the new Pi 3 got soldered together with only minimal
prompting from Phil, and it is now assembled and tested (right side in
the picture).  After considering the matter I gave the original Pi 3
combination to Phil, because that is burned-in hardware (still a
consideration with everything-on-the-cheap designs like the Pi) and
gpsd/ntpd testing by someone who is both as capable as he is and
meatspace-near enough for us to us to troubleshoot FTF should be a
very valuable thing.

(Yes, Dave, I both soldered and dremeled in the same evening.  Your
devious plan to pull me towards the hardware side of the Force is
apparently working.)

The thing of possible GPSD interest we learned is that the SKU 424254,
that no-name GPS HAT from China, has a serious flaw.  The trace that
should deliver 1PPS from the u-blox module to GPIO05 is *absent*.
This is not a simple manufacturing error; we were able to find a link
to its schematics and the trace is missing there, too.  Which is
*weird* because the vendor page advertises that export heavily.

After some scope work that would have been utterly beyond my
competence (though I am learning) Phil was able to blue-wire the board
to do the right thing. I will document this in the Microserver HOWTO
and send the vendor a nastygram that leaves them no excuse not to fix
it.

Unrelated but fun: we Frankensteined an old-school 3-button Logitech
optical trackball (best pointing device *ever*!) with an active
PS/2-to-USB adapter that had arrived on my doorstep with a shattered
case and ripped wires but the conversion microboard still usable. That
microboard now lives inside the TrackMan case, making it fully USB and
hot-swappable.

I'll publish that HOWTO as well, but not before I buy several
three-button Logitechs on e-Bay.  The device has been discontinued
for decades, but retains many fans who I expect will jump all over
an easy conversion recipe, making them harder to find.

And the Eurogame?  Ticket to Ride: Pennsylvania.  You may judge the
quality of the play and my opponents by the fact that I, the second
runner-up in last year's World Boardgaming Championship Ticket to Ride
tournament of over 300 people, came in...third.
-- 
                <a href="http://www.catb.org/~esr/";>Eric S. Raymond</a>

No one is bound to obey an unconstitutional law and no courts are bound
to enforce it.          -- 16 Am. Jur. Sec. 177 late 2d, Sec 256