[Gpsd-commit-watch] [SCM] GPSD annotated tag, release-2.35, created. release-2.35

[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-2.35 has been created
        at  a01f3dcfe6438cb47c36970996154a766e9144a4 (tag)
   tagging  583f0157119db85dff8b15795b01782331097888 (commit)
  replaces  release-2.34
 tagged by  Eric S. Raymond
        on  Mon Dec 10 20:24:21 2007 -0500

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

Chris Kuethe (229):
      we're not at 2.34 any more
      add a --enable-gpsd-user configure option.
      a better way to use ncurses, if you've got it.
      Thinking aloud:
      Fix small error in the non-functional italk driver. From Andreas Stricker
      The start of a UBX driver, from Andreas Stricker. Not yet
      eye-candy
      our testing mechanism needs as much love and attention...
      fix Id tag
      ./configure[26667]: x: not found
      Sort message classes and IDs by number.
      Try be a little smarter about probing device permissions...
      Good day to be fixing TSIP.
      Christmas Goodies: TSIP and Navcom no longer conflict.
      Christmas Goodies: a slightly functional UBX parser.
      fill in the tag for this message
      make the packet types a little quieter.
      Fix time parsing. Prompted by Ali Utku Selen in IRC
      Some logic changes:
      UBX provides a fix type of "time only" value 0x05.
      only hexdump unrecognized packets
      when a checksum fails, say what kind of packet failed.
      allow the speed search to proceed up to 115k2
      once a packet comes in, whether or not we can decode it...
      magic to make the UBX SVINFO message parse
      better tag format
      having both enable-italk and enable-itrax is dumb.
      more hacking on italk;
      tidier italk hooks. now we checksum and correctly accept/reject packets.
      remove debug goo. "XXX" stays there until i'm more confident in this code
      extract and dispatch on the correct byte of the buffer
      I prefer hex notation
      Update the iTalk entry
      more magic numbers
      more magic
      More iTalk joy.
      italk now parses satellites, time, leap seconds and position.
      Now that italk can at least give us PVT solutions...
      eye candy
      updates from Diego Berge (navcom driver author)
      This ping is not necessary here.
      Be nice to users who, for some reason, can't resolve "localhost".
      more paranoia
      remove extra '&&'
      Type. From Andreas Stricker.
      Make this consistent with the s/localhost/127.0.0.1/ transition.
      Don't do the ntpshm thing unless time is valid. Bail out if the svinfo
      A few little patches from Greg Troxel, to make the NetBSD port happier.
      Not everyone has "." in their path...
      Now that the UBX driver can produce useful fixes, enable UBX by default.
      Fix track computation in navcom driver. From Diego Berge
      don't eat the first character of the first device name...
      set numbers to always be rendered in the C locale...
      UBX and iTalk are now somewhat useful
      US Army Corps of Engineers GPS Survey guidelines.
      More updates to the driver skeleton, like sample message decoders.
      Updates from Diego Berge
      Fix to a packet-sniffer bug handling $P..
      libusb rides again!
      Fix a typo.
      more updates from Diego Berge
      more little fixes
      do not create ".1" while testing for xml processor.
      Revert the ".1" cleanup - it breaks systems using xsltproc.
      Allow floattest to compile in the srcdir != builddir case.
      Include some missing files in make dist
      Pretty up the regression tests. Allow all tests to run, and count errors.
      A nicer cleanup of the xslt processor test
      Don't run floattest when cross-compiling. Also improve the output.
      Add a pair of simpleminded utilities for capturing and replaying data on 
a pty.
      This simple utility just wraps the body of an NMEA command...
      This script is derived from gpsd.php, and generates a plot...
      sirfctl still has a use.
      Add a little bit of documentation for my crufty hack-job scripts.
      Don't ship the generate config header or a python binary module...
      Last bits to allow out-of-tree builds to work.
      Only print the spinner if stdout is a tty.
      Suppress spinner if stderr is not a tty.
      ugly hack until we can get the preprocessor to tell us the size...
      the autoconf documentation (section 5.10.2) says...
      GPSD is sensitive to data types and sizes.
      Remove test that breaks amd64.
      only open character devices read-write;
      begin squashing negative zeros.
      Whoopsy - the fallback mode is always O_RDONLY
      italk may not need this "ping"
      More updates from Diego Berge
      Move the historical italk2 code to the italk3 driver.
      More updates from Diego Berge.
      Add support for setting IP TOS and TCP options to reduce transit delay.
      make this compile again on linux.
      cached copy of the MACM spec, since the previous version 404'd.
      MID2 time is GPS, not UTC.
      add a new chip, the LS4000, found in some new inexpensive receivers.
      html cleanup
      links to a couple of sample receivers
      i wish i had an AC12
      If I were a cruel man I'd say...
      more refactoring. Move some configuration messages around a bit...
      More refactoring: 50bps data... the NAV message
      More refactoring: satellite status
      More refactoring: navigation solution
      more refactoring: geodetic navigation solution
      more refactoring: navigation parameters
      more refactoring: ublox extended navigation solution
      more refactoring: pps time
      more refactoring: error messages
      since the error or debug messages don't actually affect the session,
      validate the packets against their known lengths.
      Better error modelling.
      fix compiler warning. from Mick Durkin
      Time for some yummy crow.
      fix the test for the no configured protocols. From Mick Durkin
      This latest version of the select bug was traced down to old drivers
      got the parameters backwards. pointed out by Frank Mertens
      link to current TODO list
      Fix some typos and note that the problem can be triggered on windows, too.
      Make the nmea parsing more robust in the face of garbage and/or hostile 
input.
      protect the reconfiguration stuff with the appropriate preprocessor macro
      Much as I hate to do this, add a command line flag (-b)...
      Install gpsd_config.h like I intented to.
      missing newlines
      committed from the wrong tree, so fix the "-b" flag.
      Allow NMEA-style messages prefixed with '!' to be passed in super-raw 
mode.
      more hardware we'd like to see
      MARC (Message ARChive) is changing domain names
      Gary reported a runtime link error.
      While NMEA lines shouldn't be more than 82 characters, one of my 
receivers...
      Document the fact that the reason we accept oversized NMEA sentences is...
      more robust NMEA parsing.
      get rid of an erroneous strlen test
      fix stupid think-o. from Antonio Eugenio Burriel
      This was dealt with in my last change of 06 Jan 2007.
      Allow gpspipe to specify the remote device too.
      Handle short writes to clients.
      Remove -d -f and -p as pennance for adding -b.
      Typo
      POSIX says getmaxyx is already (void).
      too much thinking about regexes i think. hex use a-f, not a-z.
      cope with temporary failures (EAGAIN and EINTR)
      Bump version to 2.35beta - I'm waiting on a couple more test reports...
      Fixing my portability fixes
      Add pkg-config stuff to rpm specs. From Michael R. Davis
      Parrot CK3300. reported by Andy Brown on gpsd-users
      Close the network socket like we mean it.
      Neuter nmea_send.
      Neuter gpsd_write.
      don't try to write if device_readonly is set.
      avoid calling write() directly, use gpsd_write() instead.
      avoid calling gpsd_write when device_readonly is set
      use gpsd_write rather than calling write directly.
      use gpsd_write() or italk_write() rather than write().
      make tsip_write respect device_readonly
      make sirf_write respect device_readonly
      make the zodiac writer respect device_readonly
      when i say --disable-python, i mean it.
      Not quite a blog entry...
      log the ip of the client.
      roadnav is another potential consumer of gpsd data
      extraneous tag open
      another command for protocol v4, fix some angle brackets
      add another nemerix device, and a previously unknown mediatek device
      improve connection logging
      shift blame... passing in a NULL gets you a NULL back.
      Subtle but important documentation fix from Matt Roberds on gpsd-users
      undo this change i made in 4 Aug 2005 - this is a useful log message.
      generic function to get the ip of the client, as a string.
      Add my CF GPS
      Since we now have confirmation that the BC-307 works...
      Note that I now have an antaris4...
      Note that UBX works now, and post the manual for the LS-4000
      A couple of recently reported receivers
      Compiler warning cleanups.
      don't return NULL when passed a NULL pointer or a zero-length buffer;
      Add the Galileo ICD
      The shortest reported SiRF version message is now 20 bytes
      Add definitions for the timing messages
      Print the various info messages.
      actually, strlcpy would be wiser...
      pacify static checkers that would complain about fallthroughs
      u-blox AG sent me an Ataris 4 Evaluation Kit (AEK-4T). Thanks!
      suppress idle/unused channel output.
      We get the PDOP in NAV-SOL, and set the appropriate element in the fix
      an ugly little tool to send arbitrary ubx messages...
      parse the DOP message.
      Squish the spin bug.
      usage: binreplay [-d <device>] [-r <data_rate>] [-s <port_speed>] <file>
      since we now have a parser for the NAV-DOP message, use it.
      Note that some USB receivers also need -b
      Add Wintec WGM-300
      Add ublox to the list of receivers that do/might have a protocol switcher
      wrap the NTP SHM setup in the appropriate conditional
      properly terminate the output XML when aborting program
      When running a ublox receiver in timing mode...
      gpsd.h-{head,tail} must be included in the source tarball
      Add NaviLock NL-209PU and Phonix BGR6205
      Add Garmin GPS-15
      fix comment
      Sync the structure with what ntpd expects.
      add a few more devices and links
      Update links to the AC12, and my sirfstar3 receiver died.
      shuffle some links around
      I think the error modelling is fixed. Thanks to Dave (you know who you 
are)
      Fix time extraction from SiRF receivers.
      Change the SIGHUP handler.
      Add an explict license to these files.
      whitespace cleanup
      white space cleanup
      slightly refactor cgps to move the window setup code. unless jeff beats
      avoid a core dump if somehow one of these pointers is NULL
      avoid a core dump if, somehow, the session goes NULL
      Introduce not-very-documented test mode.
      Make the python module build cope with more versions of python.
      whitespace, and wrap some long lines
      Make debugging a little easier
      fixes for trimble stuff. gives better results from my lassen iq
      d'oh. thanks to my amd64 for reminding me that i need to cache the device
      Don't die if getprotobyname fails. Patch from Ulf Lamping on gpsd-dev
      indent properly
      Add the "quicktest" framework.
      forgot to add this test log
      add a "cell" mode to the skyview plotter.
      oops. was extracting azimuth and elevation in the wrong order
      missing header
      don't print NaN if the data is not available
      convert time structure elements to human-readable scale sooner...
      run all the tests, even if one fails.
      better.
      use $srcdir, rather than assuming "." in the path.

Eric S. Raymond (131):
      Explanatory comments and cleanup.
      Diego Burge's driver builds, but it messes up some regression tests...
      Some regression test problems were caused by trying to run the PTYs too 
fast.
      Oops.  Date in the gpsd.spec changelog for 2.34 was wrong.
      Fix a build bug reported by Greg Troxel.
      Chris broke the build by not including a fallback for GPSD_USER. Fix this.
      Disable Navcom by default until we figure out why it clashes with TSIP.
      One more for the to-do list.
      Translation of regress-builder into Python.
      Better documentation of regress-builder.
      And now with filtering.
      Monte Carlo compile testing now works.
      With constraints, full regress build now takes less than five hours.
      The compile-test framework has officially become smarter :-).
      regress-builder is done, until we get so many more drivers that
      Remove a now-irrelevant comment.
      Make the build degrade gracefully if ncurses is not available.
      Better documentation of prerequisites and priv-dropping behavior, as
      Degrade the build gracefully if pthreads are not present.
      Update the README with a credit for Chris.
      Alas, the attempt to factor out pthreads broke the build.
      Small fixes for the build machinery;
      Graceful degradation of the build if pthreads is  not present.
      Added a checklist for release engineers to the Hacker's Guide, and a
      Improved description of prerequisites.
      Improved the machinery for website updates.
      Trivial splint cleanup for the Garmin driver.
      Website tweaks.  tgpsd appears to be defunct.
      Tense and markup fixes.
      Documentation changes suggested by Greg Troxel.
      Cleanup after the prototype-tty change.
      Fix a typo.
      A step towards complying with Fedora package guidelines.
      Big warnings about the Blutooth hang problem.
      Note a bug reported by Anreas Volz.
      First version of internals documentation.
      Add a multiline continuation format.
      Corrections to GPSD-NG description.
      Test for ncurses based on the header, not the library.
      Regression tests have been fixed...
      Note some recent user-visible changes.
      First step of a layering fix: move the private ldisc member to gpsd.h.
      Generate gpsd.h so it will contain all relevant configure switches.
      It would help if we get the generation process right...
      More elegant gpsd.h generation.
      Cosmetic changes to gpsd.h generation.
      Matt Roberds sent these changes to make configure more explicit...
      Documentation updates.
      Make pkgconfig aware of this package.
      Fixed three minor errors discovered by Coverity scan.
      splint cleanups, including a tweak to Makefile.am...
      Back out some code that caused the eXplorist-210 regression test to hang 
hard.
      Minor splint cleanups.
      More about good software-engineering practice.
      Change return ttype of gpsd_write to be what the Garmin driver was 
expecting;
      splint cleanups, and move a static to wre it belongs in the driver 
structure.
      More splint cleanups.
      Avoid static variables.
      Pull the device_readonly variable into the global context structure.
      Get rid of another evil global.  More splint cleanups.
      Minor cleanup to remove some excedd files from dependencies.
      More splint cleanups.
      More splint cleanups and elimination of another static...
      More splint cleanups.
      Change a cast location so splint won't barf on getsl24.
      Fix an apparent bug in the Navcom driver, signed fetch for e
      Splint cleanups.
      Brag about our Coverity results.
      Minor correction to headers list.
      Lose the nonstandard license.
      Documentation and webpage typo fixes.
      Rebuild the regression tests...
      Tilman Koshnick's changes to support Garmin USB devices.
      Add a new regression test.
      Enable clean build when Python is not present.
      Hide some functions from outside the module.
      Clear the decks...
      splint cleanup.
      Include file untangling.
      More splint cleanup.
      Address a bug in the incrementing of fixcnt spotted by Mick Durkin.
      Don't require the config file for the C++ binding header.
      Matt Roberds's patch turned out to break autogen.sh.
      Mention Diego Berge's client and fix a typo.
      Text refresh.
      -T is gone.  We'll develop a better alternative after 2.35.
      Revised rules to support device removal.
      Fix a regression failure.
      Fix a help string.
      Increase the xgps window height slightly.
      Text refresh.
      Marc Balmer's souped-up xgps presentation.
      Warnings cleanup.
      Message window shouldn't clobber the GPS data frame.
      Create and use a stub set_title() function.
      splint cleanup.
      A step towards setting the window title properly?
      Put the GPS type in the window title.
      Document query form of C command.
      Enclose the GPS frame in a form so we can decorate it.
      Create decorations -- next step is to get the containment relations right.
      Regenerate the regression tests.
      Don't include the C++ binding in libgps, it causes flaky problems at
      Zero-clamp some output.
      Merge in the new resources for the modified xgps.
      Remove a large number of bogus test files...
      Updated udev rules from Mikkel L. Ellertson.
      splint cleanup.
      Typo fix.
      Eliminate naive float comparisons.
      Dammit, this typo was supposed to be fixed two commits ago.
      Rebuild some regressions, the flat-comparison bug was producing bogus 
results.
      Log the transfer mask describing what data gets moved from devices...
      One bit of the algebra in the DOP computation from satellites was 
incorrect;
      Rebuild regressions to use the new VDOP computation.
      Fix a misleading comment.
      Improved error and rpgress reporting, plus a typo fix.
      Time-equality test has to use fabs() rather than abs().
      More specific warning message.
      Make GGA report altitude with two decimal digits of precision, not just 
one.
      Trimble output still looks unstable...
      Add more write padding to banish some fluky test failures.
      Replace tcflush(..., TCIFLUSH) with ioctl(..., FIONREAD, ...)/read()...
      Separate the regression tests into stable and unstable.
      Document the new regression-test organization.
      Avoid the regression hang on OpenBSD.
      Portability tweaks.
      Updated news paragraph.
      Fix to time conversion by Petr Slansky; passes all stable regressions.
      Bump the version number.
      Update the release instructions, since we're not shipping RPMs.

Gary E. Miller (2):
      Add sanity checking to wgs84_separation().
      Add sanity checking to Garmin lat/lon values.

Jeff Francis (3):
      Removed the entire title line.
      Typo pointed out by Michael R. Davis.
      Added a table for 1PPS compatibility/claims based on ckuethe's e-mail.

Mick Durkin (8):
      Two small fixes:
      Shift a couple of lines of driver private data initialisation...
      Updated to match programs at ckuethe's previous commit.
      minor changes to pacify splint
      Allow for the situation where a fix message is generated...
      Refactoring step and fix to ntp interface.
      minor comment correction
      a waveform with a 2 second period has a frequency of 0.5 Hz

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


hooks/post-receive
-- 
GPSD