[gpsd-dev] OS X dropped. How I think about our platform priorities

[Eric S. Raymond]

I've dropped OS X from the supported platforms.  The upstream-bugs
page says:

---------------------------------------------------------------------
<h2 href="appleserial">Unknown serial-layer issues under Mac OS X</h2> 

<p>There is some unknown issue in the Mac OS X serial layer (observed
under 10.9 and 10.10) that prevents reliable use with GPSD.</p>

<p>Devices often work on the first open of a tty after reboot, only to
fail in subtle ways on subsequent opens.  The symptom is that
<code>gpsd</code> sees garbage from the device and cannot achieve
packet synchronization.</p>

<p>This bug has been observed on both USB and Bluetooth, so it is not
the same as the problem with buggy pl2303 drivers. It also survived
gpsd's move from non-blocking to blocking serial I/O.</p>

<p>This problem has forced us to drop OS X from the list of supported
platforms pending either a fix to the buggy serial layer or the 
discovery of some magic workaround.</p>
---------------------------------------------------------------------

This is as good a time as any for me to be explicit about what I think
the project's priorities should be.  If only because somebody might
point out a use case I'm missing.

In my opinion it's Android and the embedded deployments like OpenWRT
that are the really big deal these days. Certainly they completely
dominate our numbers, with firmware instances literally into the
billions.

Correspondingly, GPSD on laptop Linux and *BSDs is not as functionally
important as it used to be.  For most people most of the time Google
Maps (which has GPSD underneath) has taken over navigation, and other
location-sensitive applications are moving en masse to smartphones for
stupidly obvious reasons.

I see the BSD ports mainly as a torture test to keep the codebase
portable and clean.  Given what the *BSDs are used for (to my
knowledge, almost entirely stationary servers) GPSD on BSD only really
makes functional sense to me as a reference clock for time
service.

OS X, to me, was always more of a "because it's there" proposition 
than a serious deployment case.  Which is mildly ironic considering
that I developed GPSD's zero-configuration philosophy partly by
thinking hard about what the Mac got right.

GPSD as a time provider for NTP Stratum 1 has been increasing in
importance, and it's something I want to see us do a best-in-class
job at.

I see more focus on mobile, embedded, and time service in our future.
My main motivation for getting 3.12 out the door now was to reduce
power drain in mobile deployments, and that objective will continue
to loom large in my thinking.

Any of you who think any of this is misjudged should argue.  I think I
would find that enlightening.
-- 
                <a href="http://www.catb.org/~esr/";>Eric S. Raymond</a>

The whole aim of practical politics is to keep the populace alarmed (and
hence clamorous to be led to safety) by menacing it with an endless series
of hobgoblins, all of them imaginary.           -- H.L. Mencken