State of GPSD

[Eric S. Raymond]

The relatively short interval since 3.11 shipped has been extremely productive.

We have an entire new fully supported port - Android. There have some
significant bug fixes for unusual deployment cases, most notably in the
handling of GLONASS satellites. There are new features, most notably
support for decoding AIS Type 25 and 26 messages.

Biting the bullet and bumping the libgps library version (required to
prevent false matches on certain structured AIS messages) enabled a
lot of cleanups and space optimizations that have been backed up on
such an opportunity. The per-device memory footprint has actually
decreased, which may be significant for the embedded-systems guys.

All this is half the reason I'm thinking about shipping another release
before year's end, possibly as soon as three weeks out. The other
half...

I know some of our contributors complain about releases being rushed
and too much happening at the last minute. That would change - if
there were any more reliable way than a release warning to get people
other than me to wake up and pay attention to the state of the
project.

In other words - if you want pre-release testing of new features, the
time to work on them is *now*. Not when I've announced an upcoming
release and everyone is in a tearing hurry to get their favorite fix
or feature in muy pronto.

To smooth the process a little I've put together a list of people
who have taken responsibility for various technical tasks. Here it is:

= GPSD area maintainers =

This file points at responsible submaintainers for various portions
of the GPSD code, packaging, and ports.

== Modules and drivers ==

Gary E. Miller <***@rellim.com>::
Garmin driver, NTP interface code

Reinhard Arlt <***@t-online.de>::
NMEA 2000 support

== Packaging ==

Bernd Zeimetz <***@bzed.de>::
Debian

Miroslav Lichvar <***@redhat.com>::
Red Hat

Mike Frysinger <***@gentoo.org>::
Gentoo

== Ports ==

Samuel CUELLA <***@supinfo.com>::
Android

Greg Troxel <***@ir.bbn.com>::
NetBSD, Mac OS X

Chris Kuethe <***@lapdance.mainframe.cx>::
OpenBSD

// end


If you can correct or add to this information, please do. I'm
especially interested in knowing about other port maintainers.

Currently, here is status of the four major issues I idebtified last
week:

* The Android porting issue has been solved.

* The PPS code still needs careful review. The argument over the
right way to compute offset needs to be resolved *and the resolution
documented*.

* I want to track down the problem(s) causing test failures in AIS Type 6
messages on powerpc and mips. This is blocked on me regaining
access to the Debian porterboxes, something Bernd Zeimetz needs to fix.

* The Qt build is still busted. I think I could fix this in ten
minutes if I could reproduce it. Can anyone clue me in on (a) how
to check if Qt is installed on current Ubuntu, (b) how to install
its package if it's not there, (c) the pkg-config magic required in
scons? This used to work; I need to understand what has changed...

Now I'll add a fifth:

* Roy Barkas: is tracker bug #43039 (?WATCH on one tcp connection
returns data from all tcp connections) still alive?

I'm also waiting on Gary Miller to write a better explanation of the USB bug
he thinks causes negative offsets in the Macx-1 for the Time Service
HOWTO.

If I could get all of these cleaned up, I'd ship 3.12.

[Roy Barkas]

Post by Eric S. Raymond
* Roy Barkas: is tracker bug #43039 (?WATCH on one tcp connection
returns data from all tcp connections) still alive?
I've done some testing just now and it appears that this can be closed.

I've got two tcp connections in to gpsd from two different ais sources and
the gpsd output is correctly sending the decoded output per the request.

I don't know if something or someone fixed it or if my initial report was
erroneous.

[Eric S. Raymond]

Post by Eric S. Raymond

Post by Eric S. Raymond
* Roy Barkas: is tracker bug #43039 (?WATCH on one tcp connection
returns data from all tcp connections) still alive?

I've done some testing just now and it appears that this can be closed.
I've got two tcp connections in to gpsd from two different ais sources and
the gpsd output is correctly sending the decoded output per the request.
I don't know if something or someone fixed it or if my initial report was
erroneous.

Either way, I'll take it. Thanks; marking it fixed.

[Greg Troxel]

WHy is "packaging" and "ports" different? The packaging section is
just ports for one OS.

[Greg Troxel]

Post by Eric S. Raymond
I know some of our contributors complain about releases being rushed
and too much happening at the last minute. That would change - if
there were any more reliable way than a release warning to get people
other than me to wake up and pay attention to the state of the
project.

There isn't a more reliable way, because people with too few cycles are
willing to check things for a release but not every week, and aren't
always able to do it within 48h. All I'm asking is for about 1 calendar
week between freeze announcement (which should really mean actual bug
fixes only, not random other changes that seem mostly harmless) and
tagging.

Post by Eric S. Raymond
In other words - if you want pre-release testing of new features, the
time to work on them is *now*. Not when I've announced an upcoming
release and everyone is in a tearing hurry to get their favorite fix
or feature in muy pronto.

That's a separate issue. Last-minute rushes cause bugs. So when you
announce it's release time, you could have the freeze start then.