[Gpsd-commit-watch] [SCM] GPSD annotated tag, release-2.38, created. release-2.38

[Eric S. Raymond]

This is an automated email from the git hooks/post-receive script. It was
generated because a ref change was pushed to the repository containing
the project "GPSD".

The annotated tag, release-2.38 has been created
        at  7eb07a9e8063fee5fdae9edbcd9ae431fd60e932 (tag)
   tagging  5bbc4c62d9408c581deb4c3faa54215028069fd8 (commit)
  replaces  release-2.37
 tagged by  Eric S. Raymond
        on  Tue Feb 10 15:19:23 2009 -0500

- Log -----------------------------------------------------------------
Tag for public release.

Bernd Zeimetz (6):
      Updating list of tested python2.5/libc/.. versions
      Fixing broken identation as '\t' != '    '
      Fixing more whitespace indentation bugs.
      More Python indentation fixes.
      Fixing last commit - one part was idented too much.
      Making gpsfake compatible to python2.5 again.

Chris Kuethe (122):
      Movin' right along.
      add motosend - a tool for thwacking at motorola-type receivers
      allow the SBAS diamond icons to be filled...
      Move the ZEROIZE macro to pacify fedora's compiler.
      talk a bit more about timing receivers...
      Can't use navigator devices like the Go910 or Nuvi650 with gpsd.
      Allow communication as slow as 1200bps.
      Rip out the openbsd-specific timestamping support.
      A fix to date handling.
      rtcm requires types larger than 32 bits. test for, and use 'em.
      paste-o
      emphasize the existence of non-8O1 hardware
      typo in comment
      fix broken tag
      holux m241
      change various PI macros to GPS_PI
      #include <sys/stat.h> for umask(2)
      also allow trimble packet 0x5a (raw measurements)
      accept tsip packet type 0x48 - system messages
      Initial support of Thunderbolt output...
      whitespace fixes
      Unify the header protection macro style a bit.
      also hook up ntpshm in the thunderbolt case. from Scott Mace on time-nuts.
      C syntax fix...
      rename "FILE * fd" to "FILE * fp"...
      access to unnamed unions is illegal in C99. found by pcc.
      Add another Mediatek reference. From Henk Fijnvandraat
      cast this float to int since we're printing it as int
      validate sub->fd.  From Bartos-Elekes Zsolt on gpsd-dev
      Clean up a canceled thread. From Bartos-Elekes Zsolt on gpsd-dev
      commit 5d5c12da291fda82c41c69814d7f89c8f481e2f5
      better detection of curses
      put the listen address in network byte order.
      invert the sense of "-n";
      inverting the behaviour of a flag is bad... no cookie for me.
      don't truncate "ERROR\n". from Jan-Benedict Glaw
      typo and grammar fixes. from Beat Bolli
      bunch of fixes to ubx, particularly for the ublox5. from Henk Fijnvandraat
      fix the order of merging in time data. from Cvachoucek
      avoid a crash while parsing time. from greg troxel
      fix parsing of geodetic navigation data, from Philip Derrin
      Commits from the caribbean: 20.628361N / 87.067351W
      Commits from the caribbean: if the HDOP in the Geodetic Navigation Data
      whitespace
      whitespace
      Greg Troxel's patch to allow the user to decide whether to listen on the
      fix broken markup
      Add a wrapper function around gpsd_hexdump to avoid hexdumping buffers...
      remove motif dependency, from greg troxel.
      add and document lla2ecef
      garreset: for un-shooting your gps18 in the foot. ;)
      add a simple makefile for some of these tools.
      quell unused variable warning
      fix invokation of mkgmtime_test
      fill in my library versions
      merge mkt-3301 support. from Henk Fijnvandraat
      make this compile on (at least debian) linux. from bzed
      getting closer to release time, crank the version to rc1...
      whitespace and update diego's contact info
      more whitespace
      try extract a little more information from the Geodetic
      crank up the precision of fixes to 9 decimal places.
      typo fix
      prototype processMKT3301() to silence compiler warning
      Safty fix to Zodiac initialization.
      make it possible to built a tsip-only gpsd
      better types for these fields
      eye-candy
      silence unused parameter warning
      fix format warning
      better fix for format warning
      fix use of unbounded string functions and format warnings
      use "%z" to print size_t-sized things. also, whitespace
      fix format warnings and whitespace
      fix format warning
      fix format warnings
      fix format warnings
      link navcom docs
      fix format warnings
      fix format warnings
      fix format warnings and whitespace that was making my eyes bleed
      fix format and sign comparison warnings
      bump to rc2
      expand the Id  keyword
      Pacify automake
      little fixes
      break apart the geoid test driver from the main library.
      use some of GCC's attributes...
      crank to rc3
      expand the Id keyword
      fix a bare '&' to make this parse. also, expand the UERE acronym
      hush
      hush
      prolific driver bugs are still hurting OS X users.
      hush
      let the user specify a particular LIBTOOL... mostly for OSX users
      fix reporting of a device going offline. from Michael Ledford
      silence some unused variable warnings
      send new gps reports to me and the gpsd-dev list
      format the output to resemble gpscap.ini. allow btglitch to be specified.
      missing includes
      Add some privacy notes about gps reports.
      usb devices suffer from "the bluetooth bug", too.
      small tweaking where we know about devices with broken serial interfaces
      reminder of which file to edit...
      19200 is also a valid speed
      some fixes to make this a little more find-n-replace friendly.
      more fixes to make this useful as a driver template
      more little fixes
      add some extra decoder trace messages
      spacing
      more paranoia
      driver for novatel superstar2.
      better fix suggested by Greg Troxel
      add my superstar2
      Allow separate reference documents to be specified for the chipset and 
device
      use the right vendor, make use of some of the engine-specific 
references...
      more documentation linkage
      unbreak UBX
      Add regression test load for my LEA-4T
      crank to rc4. hopefully this should be the last one.
      fix typo so this compiles again

Eric S. Raymond (301):
      Clarify that the existing RTCM code describes 2.x.
      More preparation for RTCM 3.  No behavior changes.  All regression tests 
pass.
      File renames in preparation for RTCM 3.
      Typo fix and documentation tweak and a bit of type refactoring.
      Introduce RTCM3 reporting structures.  No behavioral changes yet.
      More namespace shuffles preparing for RTCM3.
      RTCM104v3 packet detection.
      Give a partial description of RTCM 3.x.
      Add a CRC-24Q implementation.  It's wrong, but it's a start.
      Document the error polynomial.
      CRC24Q checksum works now.
      More documentation.
      Add 3.1 fields to RTCM reporting structure.
      Take account of latest IERS bulletin.
      Add missing file.
      -f option is long gone; remove the declaration.
      Fix unclean references to the packet output buffer from bits.h.
      Warn about an unlikely edge case.
      Comment out a claim that is presently untrue.
      Add information on time accuracy.
      NMore notes about failure modes on weird architectures.
      Splint got more capable, or something. Remove some overrides.
      Splint cleanup.
      More splint cleanups.
      Yet more splint cleanup. 3.1.2 really is more capable than 3.1.1, it 
seems.
      We are now splint clean.
      More preparation for RTCMv3.  No behvioral changes yet.
      Minor polishing for possible future code reuse.
      Reference for people who want to understand CRC algorithms better.
      Rudimentary RTCM104v3 support compiles and splints clean.
      Add portability warning.
      Python module recipe has to change because the packet getter requires 
crc24q.c.
      Add verbosity option to gpscat.
      Set some ignore properties to unclutter status listings.
      Set some ignore properties for cleaner listings.
      This belongs under version control.
      Fix gpscat to dump binary packets correctly.
      Small functional improvement in gpscat -t.
      Note that RTCM104V3 doesn't use the ISGPS underlayer.
      Cosmetic changes.
      Enable processing of multiple RTCM3 packets.
      Ignore-list changes.
      Packet state machine now recognizes RTCM3 packets correctly.
      splint cleanup.
      Improved debugging message.
      Update some documentation.
      Typo fix.
      Document some limitations.
      Note that the RTCM standards have leaked.
      Fix some broken markup.
      Make some search links available.
      Revert previous EOF_PACKET change, it was a bad idea.
      Edit ignore list.
      Patch suggested by Matthias Urlichs.
      Add a bit more about RTCM.
      Add to TODO feature suggested by Matthias Urlichs.
      Add reference to RTCM standards.
      Documentation tweaks.
      Remove a deprecated option.
      Clarify a point raised by Steve Franks.
      Fix thinko in comment.
      Ignore property edits.
      Refactoring step. Eliminate second argument of packet.c::packet_parse().
      Bugs can lurk indefinitely in the damnedest places.
      Avoid erroneously decrementing the input count on a transient error.
      Add a packet.c entry point (actually a macro) for testing...
      Packet getter is fixed so we can see RTCM3 packets.
      Avoid having to call the RTCM3 unpack function just to get type and 
length.
      Another step towards RTCM3 unpacking.
      Rename the little-endian bit-unpacking macros to be explicit about their 
LEness.
      Rename big-endian bit extraction macros.
      No longer necessary to define LITTLE_ENDIAN_PROTOCOL.
      Bitfield extractor seems to work.
      splint cleanup.
      splint cleanup.
      Use the new bitfield extractor for RTCM3 fields.
      Large step towards RTCM3 dumping.  This code is still buggy and 
incomplete.
      Include the GPS device in debug messages.
      Factor out the bitfield functions, fix a bug in one, add tests.
      Revised and corrected bitfield functions.
      Still more fixes for bitfield extraction.
      Unpack and dump two more RTCM3 messages.
      Add a description of the causes of fix latency.
      More about fix latency.
      Document our exposure if LEAP_SECONDS is wrong.
      Correct start of subframe.
      GCC now gets huffy about unchecked write(2) returns.  Fix some of these.
      Abort on I/O errors, also pacifying GCC 4.0 about handling write(2) 
returns.
      Throw an assertion on failed writes.
      The fix from 15 Jul ("Bugs can lurk indefinitely in the damnedest 
places.")...
      Backing down to Python 2.4...
      Back down to Python 2.4 to evade weird problems, possibly with ptys.
      Data points about the odd PTY-related bug.
      A bug report from bzed.
      Instead of wiring invocation of python2.4 into gpsfake...
      Committed so bzed can modify it.
      Scale the O reports of horizonal and vertical errors properly.
      Markup fix.
      Add regresion test for Toyaltek RGM3800.
      The recent error-scaling change means the regression tests need to be 
rebuilt.
      bash is not required.
      I hade WRITE_PAD set too low, alas.
      Ignore comments in NMEA test loads.
      Created a NEWS file, added NEWS for 2.37 and 2.38.
      Update the release checklist now that we have a NEWS file.
      Autoconfigure fixes.
      Include a link to a tech reference for the GSC3f-7879 chip.
      Add another regression test, fix a typo, and document a regression build 
recipe.
      Integrate the new NEWS file into the release procedure.
      My change to the scaling of O reports four days ago was backwards.
      Regression test and hardware entry for Blumax GPS009.
      Diego Burge's Qt client moved.
      Added report on Holux GR-239.
      Corrected SGS example.
      Use the method for testing for IEEE754 nan...
      When parsing the K response, do it fro, a copy...
      Merge code for Oceanserver compass from Stefan de Konink.
      Cleanup after the OceanServer patch.
      Ensure gps.py sets the valid field correctly.
      Clean up time field handling...
      Remove a to-do item that's been fulfilled.
      Typo fixes and clarifications for issues raised by Bruce Sutherland.
      Fix brown-paper-bag bug in time parsing.
      Fix UERE factors for the right 95% confidence level.
      V and P base errors aren't identical to the horizontal base error.
      Temporarily give up on computing UTC time from the O timestamp in Python.
      Rebuild regression checks with new UERE values.
      Fix a test load messed up by a timing error.
      Occasionally I need a reminder that Python is not LISP.
      Add CMR documentation.
      Minor documentation tweaks.
      Remove reference to a vanished script.
      This diff goes most of the way to getting us splint-clean again.
      More splint cleanup.
      Clean up gpsctl a bit before extending it.
      Refactor sending of control strings.
      A control_send method for the SiRf driver.
      splint cleanup.
      New gpsctl send-string feture is ready for test.
      Document a helper function better.
      Add Rad2Deg to gps.py.
      Proof-of-concept GPS simulator code.
      Cleaned up and simplified command parsing...
      Hex p[arsing fix by Beat Bolli.
      Add the -re option to gpsctl to echo framed control strings to stdout.
      Documentation fixes for NMEA parsing.
      Add GLL reporter to NMEA class.
      Generalize the -e switch so it can dump mode switch and speed switch 
strings.
      Fix fatal typo in test main.
      A bit of bulletproofing for the Python side.
      The setup file pertains to the whole package.
      Update the list of packet type defines.  Plain RTCM_PACKET is gone.
      There is no more EOF_PACKET type, so fix gpscat.
      Start on a second Python extension module...
      gpssim uses the new geodetic-separation wrapper.
      Prevent shipping of stale data as the first fix when a device is 
reallocated.
      Document the fact that cps_close() doesn't kill all threads.
      Document Robin Darroch's problem and workaround.
      Update and reorganize the upstream-bugs page.
      More data on several upstream bugs.
      Sync patch from GreG Troxel, fixes minor X11 config problem.
      More website updates.
      DBUS fix patch from  Jan-Benedict Glaw <address@hidden>.
      Add a discussion of the short-write-on-Bluetooth issue.
      Don't pack some generated files in the distribution.
      A step towards making "make distcheck" work.
      Experimental code for sending hexified binary over the control channel.
      Be explicit about which write returns we're ignoring.
      Redundant with the new -c option of gpsctl.
      Add bzed's new data on the gpsfake hang.
      Another deduction about the gpsfake bug.
      tcdrain before close of a GPS, because if there's any traffic waiting in 
the
      Experimental code for gpsctl -r option.
      Quiet a compiler warning.
      the driver switcher that's coming back to bite us.
      Device subtype probe has to be gated by the readonly flag...
      gsfake hang problem is solved.
      Remove a layer of code that was unecessary...
      Add some instrumentation.
      Fix the gpsfake hang under 2.5.
      bzed's buffering fix makes it possible to cut the write padding a lot.
      Beef up instrumentation a little.
      More instrumentation.
      Make the RTCM2 dump format self-describing with a trailing sentinel line.
      Revert an inadvertent inclusion of test scaffolding.
      Change the throttled_write() logic so each sentence shipped is an atomic 
write.
      Fix instrumentation so it reports atomic chunks as well as writing them.
      Instrument gpsfake, in an attempt to track down an apparent bug in this 
module.
      Make gpsfake use the C packet-getter...
      Improve the packet instrumentation slightly.
      Fix misleading comments.
      Fix buggy termination test,
      Transfer mask dumping is not very useful since we solved the J problem.
      Add a reference.
      We can no longer count on being able to extract a packet type for gpsfake.
      Before my 12 Jan change to the throttled_write() logic...
      Fixes all deterministic errors in the regression tests.
      Increasing the close timeout to 3 appears to banish timing errors.
      Forcing each pty to drain before closing appears to be helpful.
      Simplify the regression-test driver.
      Add option to avoid noise from the baton in batch-mode testing.
      Allow the regression-test driver to operate with debugging set high.
      Documentation fix.
      Another try at defeating regression-timing bug.
      Magic-number elimination.
      Part-conversion of hardware reference page to capability database...
      More hardware entries moves to the database.
      The database now captures all data in the HTML hardware table.
      Add some sanity checks in the wrapper class, and tweak the DB so it 
passes.
      We can now generate a better hardware data dable fom the database.
      Now in brilliant color...
      Move to generating the hardware table from the master database.
      Typo fix.
      Comment fix.
      Reduce logging verbosity.
      Reintroduce the status field.  Factor out more engine entries.
      Show PPS in the hardware table.
      Add packaging field to the CGI for gathering GPS data.
      Explain the discontinued icon and link to Chris's web form.
      GPSes in the hardware table now feature Michelin rating...
      Change "car mount" -> handsfree"...
      Add some devices for which we have regression tests.
      Give the compatibility table a useful caption.
      Fix some "What was I thinking?" code.
      We compute the ubx_write() status, so we should return it.
      Teach gpsctl to ship control strings to a UBX device, and eliminate 
ubxsend.
      nmeasend and ubxsend are no more.
      Remove productions and targets no longer used.
      When attempting reset, try shipping the control string at current speed...
      sirfctl goes away.
      Delete the specialized Garmin tweak utilities.
      We no longer need send/expect.
      Remove out-of-date TODOs about Navcom and UBX druvers that now seem
      iTalk driver also seems fully functional.
      RTCM2 is no longer as interesting.
      Make the regression tester clean up after itself better.
      Rename some files to make the RTCM version level more explicit.
      Tightening up the timing on the regressions a bit doesn't break them.
      Splint cleanup.
      Splint cleanup and removal of some unneeded headers.
      Splint cleanup. All regression tests pass.
      More splint cleanup.
      Yet more splint cleanup.
      Documentation fix.
      Fix some upload rules.
      Bluetooth bug doesn't hit USB devices.
      Add GPSDproxy link.
      Include the -b glitch in the hardware table.
      Typo fix.
      Enhance gpsctl output so it shows a capability table.
      Send mode-change strings unconditionally.
      Make a message more explicit.
      Clean up the mode-switcher fix.
      Make mode-switching to garmin binary work.
      Another correction for the mode-changer code.
      Magic-number elimination.
      Fix mangled mid27 display.
      Typo fix.
      Explain why the early revision logs look so broken.
      Die gracefully if input cannot be packetized.
      Splint cleanup.
      Tweak display so 't' command cleans up after itself as it should.
      String typo fix.
      Give gpsctl the ability to do -c to a Garmin device.
      Define a control-sender message for Navcom devices...
      Make gpsctl list speed switchers.
      Supply a control-send method for TSIP devices.  Not yet tested.
      Update the informal supported-devices list on the GPSD man page.
      splint cleanup.
      Refactor so we can re-use the hex dumper.
      Make hex_escapes() generally available.
      Revert a mistaken commit.
      The 'g' mode switch command now requires, and returns, 'rtcm104v2'...
      sirfmon can read from gpsd again.
      First step towards gpsmon -- sirfmon now uses generic packet-getter.
      Document otherwise mysterious code.
      Don't ship a mode-change string with a zero baud rate in it...
      Mode set fixed.
      Make gpsctl -n work reliably on SiRF devices...
      Make gpscat robust against an initial bad packet fragment from a tty.
      splint cleanup.
      Add -D option to sirfmon.
      Begin refactoring to use the session structure fully.
      Refactoring step.
      splint cleanup.
      Make the tool time out if it can't open the device.
      Improved error messages.
      Document a limitation.
      Set hex-dump level properly in gpsctl.
      Another gpsctl example.
      Explanatory comments and one magic-number elimination.
      Eliminate anticipatory driver switching.
      Add a packet_type member to the device structure.
      Explain why we wait for packets.
      Suppress probing for subtype if we're just sampling packets.
      More thorough suppression of subtype probes.
      Instrument autodetection a little better.
      CP210X rule suggested by bzed.
      Add whitespace for clarity.
      Ready to ship external release 2.38.
      Make the FTP upload production actually work.

Greg Troxel (4):
      Explain why TESTS=testregress doesn't work and what we might do.
      In addition to Xaw and Xaw8, also look for and accept Xaw7.
      Explain how to run regression tests, and that one don't has to install
      Look for Xaw8 first, then Xaw7, then Xaw.

Jeff Francis (14):
      I've been meaning to add this for a year or two...
      Added Garmin GPS-17HVS per user report
      Add options to gpspipe to run as a daemon...
      Oops.  Daemonize *then* sleep...  Duh.
      Everyone should share equally in the joy of reading the gpspipe output 
files.
      My bad.  I let something local slip into the "real" tree with the last 
commit.
      Add the option of displaying calculated magnetic heading rather than
      typo
      This value is out of the realm of believability...
      Added new receiver info per user report.
      Cleaned up gpsdlcdd.c a bit and moved it into the "real" code tree.
      Some wise changes from ckuethe.
      Fixed heading.
      Fixed garmin mode change to NMEA.

-----------------------------------------------------------------------


hooks/post-receive
-- 
GPSD