[gpsd-dev] Broken gps.h got by me - snap release of 3.13 within 48 hours

[Eric S. Raymond]

I do very frequent fine-grained releases on all my other projects -
hey, didn't some guy write a book about that once? - but the GPSD
codebase has resisted this practice successfully in the past.

That's going to have to change at least once.

Some well-meaning person partitioned some port cruft off into a new file
compiler.h, and I missed the fact that it had been included in gps.h.

This is bad; gps.h has to be standalone, because it's installed in
/usr/include for the use of people writing C clients with libgps.
With a 3.12 installation, such builds will break.

Fortunately - very fortunately - we've had no destabilizing changes in
core code since the 3.12 release four days ago.  The only change that
has touched the daemon other than one type rename is the addition of
the TOFF message, and that is nicely self-contained.  Most of the
action has been in gpsmon and my new ntpmon tool.

I plan to issue a snap release almost immediately.  I had been
thinking about doing that anyway one I got the NTP stripchart display
working, as an experiment to see if shorter release cycles will
work for GPSD.

Before 3.13 I'm going to try tweaking compiler.h so memory_barrier()
has an implementation that works under clang.  This is mainly because
I owe Harlan Stenn an NTP patch that should include this code and it
would be silly to qualify it twice.  Otherwise I'm done.

Get your bug fixes in now; mail saying "I got nothin'" would also
be appreciated as it might allow me to speed up the release.
-- 
                <a href="http://www.catb.org/~esr/";>Eric S. Raymond</a>

"The state calls its own violence `law', but that of the individual `crime'"
        -- Max Stirner