[gpsd-dev] Progress is being made towards a GPSD release

[Eric S. Raymond]

Select-buzz suppression is back on the feature list for the 3.12
release.  I've been able to seriously simplify and improve the low-
level tty handling; it runs in blocking mode without even a VTIME set
now, which is excellent.  Device I/O will now present to the select()
in whatever chunks the tty layer assembles rather than as single
characters.  If this doesn't solve the problem, we have the option
of adaptively setting VTIME when the loop is spinning too fast.

Gary and I have been pounding on gpsmon and it is in much better shape.
PPS bars are now reliable in both direct and client mode, and the PPS
offset display looks right (e.g. matching ntpq).

Currently known bugs:

Hal Murray reports scons check dies trying to install gpsfake. Not
reproducing elsewhere.

Hal Murray reports  gpspipe -w dies with "Terminated".  It's working
for me; Hal, can you be more specific about the circumstances?

Gary reports ntpd/JSON doesn't work for him, and seems to have pinned
this to some initialization problems in ntp.  I don't consider this
a release blocker.

Hal Murray & Greg Troxel: please re-check the WRITE_PAD and
CLOSE_DELAY values you reported in gps/fake.py.  The device-I/O
changes mean they may no longer be correct.

Frank Nicholas reports total test failure on Gentoo.  Why is unknown,
but I have seen something like that on Debian porterboxes built 
without pty support.
--
                                                        >>esr>>