[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[gpsd-commit-watch] [SCM] GPSD annotated tag, release-2.90, created. rel

From: Eric S. Raymond
Subject: [gpsd-commit-watch] [SCM] GPSD annotated tag, release-2.90, created. release-2.90
Date: Mon, 03 Oct 2011 11:35:58 +0000

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-2.90 has been created
        at  db740b5460435648d2c3d1011786173fec35a346 (tag)
   tagging  b51f606e6ee2bb0da2199ba6b8f7ad4cac472146 (commit)
  replaces  release-2.39
 tagged by  Eric S. Raymond
        on  Fri Dec 4 15:50:01 2009 -0500

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

Bernd Zeimetz (26):
      Install gpscat via require gpscat.1 to build. install gpscat manpage.
      Include gpsd.hotplug.wrapper in EXTRA_DIST.
      Allow to disable the hotplug wrapper.
      Move implementation of constructor to .cpp file, initialize gps.
      gpsctl.xml: Typo fixes Reindent with 4 spaces per level Simplify response parsing Fix C response parsing Fix gpsdata.__repr__() Allow host and port command line parameters Improve setting the valid bits in O responses Unknown data is excluded before the switch Make poll() read all buffered data before returning
      ac_python=no when --disable-python is set.
      In gpsd.hotplug.wrapper also check if gpsd is started automatically.
      Several fixes for the Debian gpsd init script.
      Adding udev rules for u-blox AG, u-blox 5 and similar models.
      Remove bashisms from gpsd.hotplug.wrapper and use /bin/sh.
      Don't check for "$START_DAEMON" = "true" in gpsd.hotplug.wrapper anymore.
      Use __file__ as location of the holtplug script.
      Use environment settings passed from the hitplug wrapper script in 
      Updating packaging/etc files from Debian
      A missing % in gpsd.hotplug caused USB autostart to fail.
      gpsd.hotplug.wrapper needs to export environment settings for 

Chris Kuethe (110):
      make the ubx monitor useful
      update diego's contact information
      Bye-bye gpsflash.
      update some urls
      add some notes about raw measurements
      Fix for incorrect length.
      Fix typo. From Matt Roberds
      Fix number of data bits. From Matt Roberds
      remove some debug formatting
      make sure trimble devices are set up correctly. from openwrt
      Add some hooks to start raw data processing
      change the signal strength member to float.
      fix display
      add reported devices from the mailing lists
      make the output be closer to what we need in gpscap.ini
      accept trimble raw measurements
      Fix a segfault in the Zodiac driver.
      Admit to another crash being recently discovered. Suggest the use of
      rename PSEUDORANGE_SET to RAW_SET...
      Superstar2 can now stash raw measurements -
      more TSIP fixes
      bunch o' changes that make my superstar a lot happier.
      fix some protocol detection.
      schedule output of gps system message. not supported on all receivers 
      fix for bug 15795 - char devname[] shadows a standard function
      Make this work with "use strict". also, make it generate mime attachments.
      ... and of course i forget to change the destination address back.
      how could i have possibly screwed up superstar2 detection so badly?
      make gpspipe exit rather than spinning if the server goes away.
      remove ugly hacks that are no longer needed now that i fixed the sniffer,
      take a hint from the tsip driver...
      pull a couple of choice values out of the ephemeris and iono messages.
      whitespace cleanup while i think about signed-vs-unsigned compare warnings
      clean some whitespace and quell compiler warnings
      quiet compiler warning
      create a net_ namespace for gpsd-as-a-client protocols.  currently these 
      missing header, and whitespace
      change the dgnss_ prefix to netgnss_
      strange. somehow HAVE_SYS_STAT_H was being defined
      GPSD can now be a gpsd client
      whoopsy, forgot to crank the leap seconds.
      allow make dist to work again
      UNUSED is not something that config.h needs to define. move it to gpsd.h
      actually, UNUSED is better for our purposes in gps.h
      sync with move of UNUSED
      Javad has some new hotness, link it.
      I don't think a singular LOS matrix is worth a LOG_WARN.
      Y message can have more than 12 satellites. Reported by Toget on IRC.
      pass raw subframes to the raw subframe decoder
      currently unused, extract the source prn and channel.
      API change: callers of the subframe decoder are now responsible...
      italk can do raw subframe output too
      a little bit of handy glue for network transported GNSS data
      correct word-mangling magic.
      fallout from API change. drivers need to do some basic subframe 
      system call failure should be tested against -1, rather than "< 0"
      don't return -1 if ((errno == EAGAIN) || (errno == EINTR));
      refactor the output routines.
      bump major version of libgpsd:
      avoid use of unbounded strcpy
      avoid compiler warning about unused args
      initialize a couple of variables to silence a compiler warning on netbsd
      print (s)?size_t variables with a %z format string
      fix a bunch of things so this can be used as a driver skeleton again.
      Rough oncore driver.
      SVN Id tags
      Stuff I've hacked on lately
      catch up with changing the signal strength to double.
      Make the oncore sniffer use the known lengths of packets.
      forgot to commit this bit.
      move the oncore packet length tester to where it really should be.
      couple of little fixes from Hakan Johansson
      Oncore driver for gpsmon, from Hakan Johansson
      typo, from Hakan Johansson
      Connect the oncore decoder to gpsmon, and fix some display errors.
      More patches by Hakan Johansson
      don't shadow the devname() stdlib function
      more fixes from Hakan Johansson
      Hakan Johansson made the oncore driver acutally useful.
      no strcpy.
      we have replylen - use it to bound this copy. duh.
      Rename the function to oncore_payload_cksum_length()...
      fix prototype of aivdm_dump
      silence complaint about unused var and missing prototype
      create a macro NITEMS() rather than using 
      gps_json.c:358: warning: sizeof(pointer) possibly incorrect in argument 2
      Add a bit of detail about the ND100 and its buggy firmware.
      missing $srcdir
      pacify splint
      respect the $Id$ svn keyword
      be a little more tolerant of acceptable packet size
      cycle reporting for the tsip driver
      whitespace, and fix the printing of empty masks
      Yet Another attempt at getting all the code to build without an xslt 
      mention -G in the usage line too.
      accept and parse lines without a timestamp
      improved TSIP validation, from Chris Adams
      Fix typo. "MKT" should be "MTK", etc. Prompted by heeen on IRC
      merge in the rest of chris adams' SVeeSix patch
      missed a couple in the MKT / MTK rename
      fix the probe logic, thereby unbreaking the rest of the protocols.
      these speed switches are no longer necessary; tsip can probe for parity
      flush the input buffer if it gets too full and nothing is detected.

Eric S. Raymond (973):
      Commits are open again.
      Change GPSD-NG introducer to '?'.
      Mac OS X library bug fixed.
      Restore wait-for connect as an accessible mode and the default.
      First cut at implementing Matthias Urlich's suggestion about a release 
      Delayed device close is working.
      Autotools fixes from Richard Hansen:
      This patch makes several changes related to the man pages:
      Python build patches by Richard Hansen ("I" in this comment):
      Garmintxt integration patches from Petr Slansky.
      Typo fix.
      Fix patch for Python 2.4 support from Richard Hansen.
      Add geoclue to the Related Resources.
      Pacify splint.
      Slightly modified version of another Richard Hansen patch to cope
      Remove deprecation warning on -n, dev list traffic shows the power
      Get a size_t format right.
      Even more delayed allocation on wait-for-connect.
      Cleanup by Richard Hansen; make the multi-output hack less ugly.
      Improve the Bluetooth explanation.
      More typo fixes and corrections.
      Splint and compiler-warning cleanup.
      Document autoconf and automake levels known to work.
      Slight revision of the GPSD-NG sketch, adding a response introducer so
      Add latest version of sample NG session.
      Typo fix.
      Another typo fix.
      Markup improvements and more typo fixes.
      Typo fix.
      Support, and documentation, for AIVDM messages type 6, 7, 8, 12, 13, and 
      Comment fix.
      We now have a working regression test for AIS Type 8 messages...
      Include test_float and test_bits properly in the build machinery.
      Improved documentation and regression testing for the bitfield extractor.
      Fix dumb mistake in Repeat Indicator extraction.
      Regression test for message type 6 passes.
      Fixed equally dumb bug, getting fragment number from the wrong field.
      Restore correct regression test for Type 5.
      Significant simplification of the packet-getter interface.
      Python interface needs to track that last change, too.
      Explain a tricky issue that might bite us someday.
      Type 6 regression test is correct.
      Workaround for the Garmin configure-loop problem.
      Correct the documentation (and, I hope, implementation) of messages 7 and 
      Minor correction.
      Better explanatory comments.
      Minor splint cleanups.
      Bartos-Elekes Zsolt:
      Patch by Petr Slansky to avoid both splint and compiler warnings.
      Update the news.
      Add explicit six-bit-ASCII decoding table.
      Add table that is explicit about ASCII-armoring.
      Bump AIVDM doc to version 1.4.
      List obsolete sentences.
      Add endianness clarifications.
      Small build patch by Luotao Fu.
      Don't clobber the destination of gpsd_hexpack; thanks to  Joshua
      Correct intepretation of ASCII-armored field 7.
      Prevent a possible buffer overflow from a malicious packet.
      Correct AIVDM message type 5 interpretation...
      Patch suggested by Micheal Tandy...
      Fix for incorrect hex-transmission code.
      Typo fix.
      Markup and typo fixes.
      Added another AIVDM regression-test example.
      Comment fix.
      First cut at handling AIS message 19.
      Add regression test for AIS Type 19 message.
      More consistent logging for AIS sentences.
      Teach the AIVDM driver to handle message type 11.
      Note some ITU 1371-3 changes.
      Add support for decoding AIS message type 10.
      Typo fix and changelog update.
      Add logging for Type 10 messages.
      Describe message type 21. Add section on improving this document.
      Typo fix.
      Add a regression test for AIS type 11 (expected to succeed cleanly)...
      Add information on Message Type 24 and ITU-1371-3 flags in Message Type 
      First cut at support for AIS Type 21 messages (not tested).
      We now correcttly interpret and dump AIS Type 24 messages.
      Correct the test for auxiliary-craft MMSI.
      Add support for the AIS Version 3 flags in message type 18.
      Clean up splint and compilation warnings, fix minor gpsdecode option bugs.
      AIS documentation fixes.
      Add info about JSON-AIS.
      The emailed report from the CGI didn't include the model number!
      Abandoning the gpsflash project...
      Clarivy what ITU-1371-3 is.
      Splint cleanup.
      First cut at JSON dumping of AIVDM.
      Second phase of JSON conversion; get string quoting right.
      Another step towards JSON-AIS output.
      Chasnge hex fields to integer in JSON-AIS output.
      Update regression test to reflex new hex output format.
      Fix JSON output format for packed binary string data.
      JSON-related name cleanup.
      Typo fixes and minor clarifications.
      Regularize names so the C structure members and JSON fields match up.
      Cope with broken \r\r\n packet trailers from the Jackson Labs Firefly-1a.
      Typo fix in comment.
      Add member names for the bitfields.
      More JSON cleanup.
      Stop extracting and dumping spare bits from AIS message fields.
      Also remove the "spare" member name from the spec.
      Go back to passing AIS out-of-band values as-is into the JSON dump;
      Minor corrections to JSON dump code.
      Comment fixes.
      Markup fix.
      Fossil removal.
      Explain CNB decoding better.
      Describe message AIS types 15-17.
      Splint cleanup.
      Whitespace cleanup.
      Typo fix.
      Teach AIVDM driver to handle AIS messages 15, 16, 17.  Not yet tested.
      Splint cleanup.
      Suppress a spurious memory-leak indication from splint.
      Fix brown-paper-bag error in JSON formatting.
      Add descriptions of AIS message types 20 and 22.
      Dump Navigation Aid types as strings..  Fix an attribute name.
      Add support for parsing and dumping AIS message types 20 and 22.
      Correct some JSON attribute names.
      Typo fix.
      Dump common navigation block more sensibly when speed is not available.
      Typo fix. Regression tests pass.
      AIS member name corrections, new details on out-of-band values from IALA.
      Another step towards pure-Python AIS decoding.
      Fix a slightly broken format string.
      Message 6 in the Python decoder.
      Typo fix.
      AIS messages 7 and 8 in pure Python.
      Correct processing of AIS message 9.
      Make the packet getter a bit more explicit about checksum failures.
      Type 9 AIS message in pure Python.
      Message 4 latitude and lomngitude were being dumped in the wrong order.
      AIS message stypes 12-14 in pure Python.
      AIS Message types 15-19 in Python.
      Use validator hooks a bit more systematically.
      AIS message 20 in pure Python, and a lot of explanatory comments.
      AIS Type 21 in pure Python.
      Implement scaling flag and an output format resembling Kurt Schwehr's.
      AIS message type 22 in pure Python.
      All known AIS message types are now handled in pure Python.
      Documents more out-of-band values and treat radio status blocks more 
      Document better what is tested and what is not.
      Package packet-fragment assembly as a generator.
      Handle the wacky run-type conditional in 24 part B.
      Handle AIS Type 21 name extension field.
      Comment fixes.
      Typo fix.
      Comment fixes.
      Simplify the way field metainformation is passed forward.
      Fix handling of out-of-band values.
      Improve bit vector encapsulation.
      First step towards new protocol. Implement ?TPV and ?SAT.
      Oops, fix JSON syntax - uses Python-style dictionary notation.
      Use boolean JSON values where appropriate.
      Define a configuratrion switch for NG protocol support.
      Scratch JSON parser.  No unit test yet, that gets built next.
      Get JSON parser unit test to the point where it will at least build.
      JSON unit test verifies correct integer and string attribute parsing.
      JSON parse unit test verifies correct function for string, integer, and 
      JSON parser now verified to handle booleans.
      Verify integer and real defaulting in JSON.
      First steps towards parsing JSON arrays.
      Fall back to working version of JSON regression test.
      Fix a compiler warning and change a prototype...
      Stub JSON array parsing. Unit test passes.
      Encapsulate type information about JSON arrays.
      Use properly bounded copies for JSON string fields.
      Make #defines for JSON error codes.
      Much simpler code for filling sub-objects.
      Set up initializers for JSON satellite data parsing.
      Use double rather than float for satellite signal strengths.
      Subarray parsing in JSON is working, now to tighten up the validity 
      Fix booleans potion of JSON unit test.
      Tighten up the JSON-parsing unit test.
      Add entry point for dumping JSON parse error condition text messages.
      Comment typo fixes.
      JSON parser can now handle arrays of strings.
      JSON error-handling cleanup.
      Refactor JSON handling so more code can be shared by client and daemon.
      More JSON refactoring - prepare the client libary to use it.
      Code for reading JSON TPVs is now complete and unit-tested.
      Beef up the unit test slightly.
      Fix buglet in control notifications.
      Reverse the breakout of gpsd_output.
      Add AIS as a supported type for the 'G' command.
      Un-break the build.
      A step twards reporting AIS data in gpsd.  All regression tests pass.
      Now reporting AIS packets when raw mode is set on an AIS device.
      gpsd can now dump decoded AIS data from an AIVDM source.
      AIS dumping is now under the control of the watcher variable.
      Remove ! introducer for responses.
      Add and document the device_class member, so we can do ?DEVICES properly.
      Fix the ?DEVICES response.
      Refactoring step.
      Fix some JSON generation issues.
      Cleaner implementation of ?DEVICES.
      Enhance ?DEVICES to report driver type and subtype.
      'G' command no longer has side effects.
      Make it possible to compile without support for the old protocol.
      Further separate old-style from new-style code.
      Logic for parsing ?WATCH command.
      Cleanup for the multiple-services world.
      Isolate the 'tied' structure member.
      Break the subscriber block into two structures.
      Another refactoring step.
      Rename an array for clarity.
      Separate channel control blocks from subscriber blocks.
      More refactoring towards allowing a single user to listen to multiple 
      Remove the (never-deployed) RTCM104_SERVICE code.
      A variable and type rename.
      Refactor to get rid of more USER_CHANNEL calls.
      Eliminate more USER_CHANNEL calls.
      More separation of channel objects from subscriber and device objects.
      More elimination of USER_CHANNEL calls.
      More the raw flag to the subscriber structure.
      Remove another USER_CHANNEL dependency.
      Get rid of more USER_CHANNEL calls.
      Rename a variable.
      Yet more refactoring - down to just one USER_CHANNEL() call now.
      First version of code to allow clients to listen to multiple channels.
      Implement device forcing.
      Finish device-forcing implementation.
      New-style WATCH command is working...
      Update client library to read current JSON - not yet tested.
      Thinko fix.
      Allow gpsfake to accept multiple logfiles.
      Handle some line-termination issues more cleanly.
      Pull some definitions into gpsd.c that were too widely exposed.
      Silence some compiler warnings.
      Make boolean defaults work properly in the WATCH command.
      Implemented ?VERSION.
      Improve some progress messages...
      Fix an odd and subtle bug...
      New-style notifications on device status changes.
      Refactor code for parsing WATCH responses...
      No longer necessary to document GPSD-NG as though it's in the distant 
      Added the protocol-evolution paper.
      Finished 1.0 version of protocol evolution paper.
      Typo and title fix.
      Typo and markup fixes, and two added bits of example explanation.
      More typo tweaks.
      Move 'raw' to being per-channel policy again.
      Break per-channel configuration bits out into a separate config 
      Explain the exhaustion of old-protocol namespace in more detail...
      In gpsfake, only send w+r to the slave daemon in pipe mode.
      CONFIGCHAN parsing is working and documented.
      CONFIGCHAN now works for setting the raw bit.
      JSON parsing and dumping for device configuration.
      Major steps towards CONFIGDEV. (Not yet working.)
      Make CONFIGDEV reporting work, and document it.
      Experimental 'scaled' attribute for AIS reporting; needs testing.
      Dump AIS MMSI as a string...
      Output fixes from Jan Veninga.
      Prevent SEGV on F when channel assignment fails.
      Typo fixes.
      Add some discussion of paths not taken.
      Abolish device classes.
      Added a link to the protocol-design paper.
      Property changes to track binary renames.
      Always sniff from devices on startup.
      Paul Fox's fix for the F command.
      Add regression test and gpscap.ini data for NL402U,
      Add entry for NaviSys GR300.
      CGI and gpscap.ini drifted out of sync a bit.
      A crude mode switch for the TSIP driver, by Matt Roberds.
      reference -> techdoc
      Clean up some nits in the documentation directory.
      GPSD-NG commands without object arguments now require an ending ';' 
      Argument-less commands now have a delimiter.
      Remove an obsolete sentence.
      Make the submitter field a unform feature in gpscap records...
      Code simplification.
      A refactoring step.  All regression tests pass.
      Multiple GPSD-NG commands per line are now supported.
      Clean up NG command parsing.  Only require one introducer per request 
      Channel assignment failure isn't a crash landing when done from WATCH,
      Support JSON dumping for RTCM2 messages.
      Typo fix.
      Comment fix.
      Typo fix.
      Add some constants (not yet used, but it's good to stay in sync).
      Code simplification.
      More code simplification.
      Fix progress message.
      Remove a superfluous write.
      Suppress garbage write on channel assignment.
      Back out recent breakage in the Python binding that didn't work.
      Fix the regression-test driver and the tests.
      Simplify the regression-test driver a bit...
      Don't let error messages step on other progress indications.
      gpsd_poll() now asserts the DEVICE_SET bit...
      Big simplification of GPSD_NG interface.
      Skip unallocated devices when watching.
      Prepare to switch regression tests over to using new protocol.
      Switch regression tests and gpsfake over to new protocol.
      Add code for changing modes in devconfig.
      Add cycle-time reporting and setting to new protocol.
      Ensure that new-style watchers see new devices as they are added.
      Tighten up the logic for closing devices.
      Document an advantage of the new protocol.
      Step one in making the client library speak new protocol.
      Major step towards GPSD-NG support in the client library.
      Set validity masks properly on the client side.
      Halfway to parsing ?DEVICES.
      Delimit a version report field properly.
      More steps towards ?DEVICES decoding - refactor code.
      Make the two different kinds of DEVICE notation conformable.
      Add check-string attributes to JSON parser.
      More work on client-side JSON.
      Correct an example.
      Typo fix.
      Move some struct gpsdata_t members around to get better space economy.
      Cleanups made possible by last refactoring.
      Checkpoint before giving the JSON parses a notion of enumerated types.
      Give the JSON parser the ability to map strings to enumerated types.
      Library parse code, and unit test, for DEVICE responses.
      Change enumerated array JSON type to yield a bitmask rather than an array.
      Removed unused member.
      Hide an argument that doesn't need to be exposed.
      JSON parser can now unpack JSON lists into structure arrays.
      Checkpoint before simplifying JSON parser.
      Change some flag names.
      Ship device flags as integer rather than string list.
      Typo fix and removal of dead structure.
      Remove enumerated-flags feature from JSON parser.
      Update an explanatory comment.
      Parsing of JSON VERSION response, with unit test.
      Updated techdoc link, thanks to Rick Richardson.
      Things that aren't macros should not be named like macros.
      Checkpoint before major surgery on gps.h.  All regression tests pass.
      Introduce gps_stream() in the client API...
      Here's the start of the protocol-transition document.
      More for the transition document.
      Explain differences in new protocol's theory of operation.
      Make gps_stream do the right thing when a raw hook is set.
      Add flags argument to gps_stream().
      Major rearrangement of device members in struct gpsdata_t...
      Merge two structures that didn't need to exist separately.
      Parse CONFIGDEV responses.
      Parse WATCH responses in the client library.  All regression tests pass.
      Update dump of union member sizes.
      Correctly read JSON ERROR responses in the client library.
      Arrange for clients to see a VERSION response when they first connect.
      Fix a small but deadly bug in the allocation machinery.
      Markup fix.
      Website fixes.
      Banish fatal error in xgps.
      Device filtering implemented for new protocol.
      xgps works with new protocol.
      Regressions needed to be rebuilt due to a bug in the SKY dump function.
      Make xgpsspeed protocol-independent with gps_stream().
      Introduce and document optional "enable" in ?WATCH.
      Filter DEVICE lines out of regression tests.
      Remove inadvertent DEVICEs lines from check logs.
      Regression tests needed to be rebuilt...
      Make more of the test clients protocol-neutral.  Document why xgps isn't.
      Update the transition document for new gps_stream() arguments.
      Avoid reporting garbage times when you have no fix.
      Leave devices in the pool when they're deactivated.
      Improved debugging machinery for JSON parsing.
      Fix up JSON unit test.
      Fix up parsing of ?DEVICES response.
      Change the way JSON parse templates are declared...
      Follow through on previous change in the client code.
      Add character type to JSON parser.
      Unify CONFIGDEV and DEVICE responses.
      Exclude more junk from regression check logs.
      Simplify the implementation of ?WATCH.
      Re-enable older behavior that devices aren't sniffed until a client 
      Fix minor bug introdued by previous commit.
      Direct some diagnostics to standard error.
      Give gpsfake a proper main sequence.
      This may fix the race window in the regression tests.
      Fix a race condition in the test framework.
      Adjusts tests after fix of a bug in the fixbuffer-clearing logic
      Documentation updates.
      Eliminate a couple more gps_query() uses.
      Add gps_errstr() to the client API so client code doesn't have to
      Add a salutary warning.
      Miroslav Lichvar <address@hidden> had this to say:
      Update the TODO list.
      GNU sed doesn't like having no command-line arguments.
      Stop trying to use enums in the RTCM2 structure...
      Fix an implicit-declaration glitch.
      Simplify some code by making a union anonymous.
      Add an unsigned integer field type to the JSON parser (needed for RTCM2).
      First step towards RTCM2 JSON parsing.
      Parse JSON for RTCM2 message types 1 and 9.
      Correct an erroneous type declration.
      Integrate RTCM2 JSON parsing into the build.
      Banish some compiler warnings.
      Checkpoint before trying to factor out some JSON attribute-initialization 
      Create a convenience macro for JSON parser template declarations.
      Use STRUCTARRAY everywhere it's apprpriate.  JSON regression tests pass.
      Fix a minor RTCM regression...
      Parse RTCM2 message type 5, and fix an error in the RTCM2 dump code.
      Don'tr report list lengths, the parser can count.
      Parse RTCM2 message type 7 JSON.
      Give gpsdecode the capability to encode from RTCM2 JSON.  Not yet tested.
      Allow RTCM2 encoder to ignore comments.
      Make sure there's a message 1 example in the Sager test load.
      Remove parser for Sager dump format.
      RTCM2 JSON parser is feature-complete and ready for test.
      Passthrough check of RTCM2 JSON decoding works for message types 1 and 9.
      Make a regression test clean up after itself better.
      Allow mapping of strings to enumerated values in JSON.
      Refactor so all the address computation in the JSON parser is done
      Use enumerated string values for the "system" field in RTCM2 type 4 
      Update some documentation.
      JSON decoding for RTCM2 types 3, 7, and 16 is verified.
      Optimize, and add some debugging.
      RTCM2 JSON decoiding is verified and has a full regression test.
      Regularize some JSON attribute names.
      Start on client-side parsing of AIS JSON.  All regression tests pass.
      AIS message types 1, 2 and 3 JSON decoding.
      Change the check file in the AIVDM regression test to use unscaled CSV.
      New CSV AIS dump code is verified correct.
      Dump boolean fields in AIS as JSON booleans.
      AIS dumper's signature has changed.  Integrate AIS parsing on the library 
      M Joonas Pihlaja's cleanup patches for XML-Docbook masters.
      Write-lock gpsd.h when it's generated...
      Fix minor build bug introduced by previous commit.
      Prevent a core dump and spurious display of blank data.
      Prevent some SEGVs on ill-formed comands.
      Typo fix.
      Recognize and process NMEA packets while processing AIVDM.
      Create a way not to step on defaults.
      Simpligy the JSON parser to fix a bug.
      Address Debian bug #542384...
      Make the test instructions speak new protocol.
      Update the test instructions some more.
      Improved error messages.
      Simplify and regularize command parsing.
      Finally give ?TPV; and ?SKY; the axe.
      More updates to the protocol-evolution document.
      Avoid a SEGV.
      Clean up device-state dumping.
      In new protocol, don't report device information on each channel 
      On the other hand, watchers *should* get notified when the device pool 
      Make the JSON parser instrumentation less obtrusive,
      Implement rate-switching in new protocol (not tested).
      Various minor fixes for AIS dumping from Jan Veninga.
      Improve parser debugging machinery.
      JSON parser debugging.
      Make string fields that might contain double quotes safe for JSON output.
      Back out a change that broke regression tests. All tests pass.
      Arrange for double quotes in JSON string fields emitted by the daemon...
      Refactor to make each dump function responsible for its own \r\n.
      Note that GPGBS is in NMEA 01083 3.0.
      Fix calls to json_stringify to not rely on static storage.
      Fix up some line terminations.
      regress-driver -n can't be made to work, so get rid of it.
      Added mixed AIS/GPS testload.
      Fix bug in RTCM2 dumping.  All regression tests pass.
      Display both driver and subtype when we can.
      Refresh NEWS and website to reflect support for AIS devices.
      Updated FAQ for new protocol and capabilities.
      Added stream() entry point to C++ client library.
      Add a hotplug icon to the capability chart.
      Typo and markup fixes.
      Proof of concept for JSON parder generation.
      Split into spec translation and generation.
      Generator now works for messages 1, 2. 3, and 4.
      More code-generation machinery.
      Field type data for AIS message 5.
      More code generation - up to AIS message 13 now.
      Fix dumping of AIS message 19 (dte was missing).
      Fix dumping of AIS message 20.
      More fixes for dumping of AIS message 19.
      Generated AIS parser code compiles. All tests pass.
      Link RTCM2 and AIVDM client-library support only when those are 
configured in.
      Comment fixes.
      Typo fix.
      Prevent adventitious doubling of regression diff lines.
      Split eph into epx and epy internally...
      New protocol now reports epx and epy separately.
      $GPGBS processing is successful at the driver level.
      Ensure we set flag masks correctly for obscure E and Q commands.
      Break the cycle-start flag out of general status mask into its own field.
      Now make it possible for a driver to signal cycle start, cycle end, ...
      AIVDM driver reliably signals end-of-cycle.
      Use CYCLE_END condition for the 2-part mal-design of AIS message 24 
      Bring the AIVDM regression test up to date.
      Rerrange some definitions for improved maintainability.
      Refactor so RTCM2 dumping is done with the other kinds in gpsd.c.
      Dead-code removal. All regression tests pass.
      End-of-cycle latch for devices that ship $GPGBS now works.
      Add annotations to several regression loads and the capability file.
      Add a note to news.
      IS8601 dump format and some defaults were slightly wrong.
      Add some minor TODO items.
      Bit-field packing for client-side parsing of AIS messages 6, 8, and 17.
      L command disabled...
      Zodiac emits a reliable end-of-cycle indication.
      Reorganize, moving a rarely-used/semi-obsolescent stuff into  
      Add some things inadvertently omitted from the repo.
      Minor cleanup.  All regression tests pass.
      Split up eph in the Python client.
      This bug was bisected on...
      Eliminate a dependency on old protocol.
      When faking NMEA, ship error estimates as NMEA-standard $GPGBS.
      Clean up properly on signal.
      Rebuild three tests, there was a typo in $GPGBS output.  Ignore RTCM3.
      Arrange for JSON-style escapes to be interpreted by the parser.
      Ignore .i files.
      Remove a duplicative storage declaration.
      Prevent spurious failure indication from shell trap statement.
      Use $(srcdir) corrrectly.  All regression tests pass.
      Move documents to GoogleCode.
      Register some developer convenience scripts (not stuff to be shipped).
      Copy the svn:ignore property from the svn repository.
      Avoid an uninitialized-variable issue.
      Uniform treatment of line terminations.
      Enable JSON parser to recognize a type mismatch.
      .html files in main directory are generated, but some in www are not.
      Added more type testing in the JSON parser.
      Reversed earlier change of MMSI to a string.
      Fix some broken AIS dump code. All regression tests pass.
      Improve synthetic test load for AIS unit test.
      Documentation and comment tweaks.
      Refactor NMEA driver so all sentence timestamp recording is done in one 
      Apparently I fat-fingered editcomment while committing it. Fixed.
      Document a $PASHR format.
      First cut at log cycle analyzer.
      Add cycle detection.
      Group events into bursts.
      Another analysis step.
      Add more analysis, and document what it does.
      Play well with devices like the Humminbird that emit $IN sentences.
      Carry timing info through all reporting stages, for better debugging.
      Make cycle_analyzer better able to handle short logs...
      Add some filtering flags.
      First cut at cycle detection from JSON.
      Note a strange phenomenon in one of the logs.
      Correct some bad packet dumping.
      Stop interpreting SiRF GND.
      Sigh, JSON can't be autodetected.
      Trying to use the Python standard library JSON parser...
      Conforming JSON libraries don't like leading zeros on numeric literals.
      Cycle analysis via JSON now works.
      Document some Zodiac behavior.
      Improve documentation.
      More documentation tweaks.
      UBX driver now signals cycle end reliably.
      Reliable cycle-start indication for almost all drivers.
      Fixed a bug in the cycle-signaling code for SiRF.
      Refactoring step.
      Minor bugfix.
      Fall back to parsing JSON out of a .chk file...
      Slightly better job of whitespace trimming.
      Add regression test for Navcom chips.
      We don't have to collect cycle-time and start-of-cycle...
      Partial splint cleanup. All regression tests pass.
      More splint cleanup. All regression tests pass.
      Yet more splint cleanup. All regression tests pass.
      Continuing the splint cleanup.
      Bride of the Splint Cleanup.  All regression tests pass.
      Son of the Splint Cleanup.
      Splint Cleanup Meets Billy The Kid.  All regression tests pass.
      Return of Splint Cleanup.
      Note that we handle AIS receivers.
      Dyke out the remotegpsd code.
      Bride Of the Son Of Splint Cleanup.
      Guard some inclusions that splint doesn't like.
      Finally found the right set of exclusions to stop splint from choking.
      Correct some splint annotations.
      Add and annotate more stuff to be splint-audited.
      NMEA end-of-cycle detector appears to be working.
      RMC and GLL emit valid time even when navigation warning is on.
      J command and switch are gone.
      On second thought, we ought to support two client bindings...
      End-of-cycle detection solves the buffering problem.
      Banish J. Bring the status flag list up to date.
      Improved handling of defaults in DEVICE command.
      Exclude various sorts of object files in git status.
      Fix a glitch in the product database.
      Don't calculate unused checksums.
      Merge the configurator and probe_subtype methods.
      The configurator member becomes event_hook.
      Fold the revert method into the generic event hook.
      Fold the probe_wakup method into the new generic lifetime-event hook.
      Rename event_revert to event_deactivate.  No behavior changes.
      TSIP driver was a bit too ambitious about setting mode from skyview;
      Never report time uncertainty if we don't have time.
      Revert incorrect part of previous commit.
      Add some needed string quotes.
      Simplify the regression-test driver.
      More simplification of the regression test driver.
      J is gone.
      Give DOPs their own structure...
      Add clear_dop().
      Begin using clear_dop to ensure that gpsd never reports stale DOP values.
      More clear_dop() calls go in place. All regression tests pass.
      Using clear_dop() means we can get rid of instances of several different 
      More simplification of DOP handling. All regression tests pass
      Give the ubx driver a single point of truth about DOPs.
      Typo fix.
      Enforcing single-point-of-truth pays off; this changes eliminates spurious
      Rebuild a regression test.
      Split libgpsd_core.c in half.
      Separate DOP flags are gone.
      Fix an ambiguity pointed out by lvh.
      Remove a variable declaration that is now unused.
      Changing a guard in gpsd_binary_dump() gets rid of a bunch of junk GSAs...
      SKY response now includes DOP reports.
      Change the driver event set to be more orthogonal.
      Guard against possible race conditions when invoking the ?DEVICE command.
      Remove a fulfilled TODO item, add some warning comments,
      Suppress excess \r\n in ?DEVICES output.
      Disentangle the packet raw hook used on the client side...
      Restore the operation of the raw switch.
      Refactoring step.
      Minor refactoring step.  All regression tests pass.
      AIVDM is a textual packet type.  Treat it uniformly as one.
      Splint cleanup.  All regression tests pass.
      Minor code rearrangement, clearing the decks for larger ones.
      Restore super-raw mode.
      Code cleanup.  All regression test pass, codebase splints clean.
      The NMEA-dump flags is now properly separated from the raw flag.
      Daemon now has its own version symbols for the wire protocol...
      Added the Future page to the website portion of the repo.
      Website polishing.
      Repair the superraw test.
      Typo fix.
      Simplify the API. The second argument of gpsd_activate() was a fossil.
      Add ckuethe's most requested JSON feature: WATCH a specified single 
device only.
      Documentation improvement.
      Make a re-entrance-friendly version of gps_open().
      Use the new gps_open_r entry point.
      Get rid of special command-shipping code.
      Old-protocol dependency removed from gpsmon.c.
      Remove a todo that makes no sense since J is gone.
      Integrate Chris's directory relativization for building Python extensions.
      RPM spec fixup.
      Use xdop and ydop for horizontal error estimates, not hdop.
      Permissions tweak.
      Ensure that the C and Python flag masks are consistent.
      Splint cleanup.
      Remove unneeded splint markup.
      Adapt the test code so we can easily test streaming.
      Zero some NaNs in reports, prevent duplicative reports...
      Fix a very minor bug in Zodiac reporting.
      Heads up, Chris! -
      Fix incorrect processing of multiple commands pointed out by mrd.
      List known driver-level issues.
      Revert a previous simplification.
      Relativize a number of testfile creations properly.
      More test-production fixes.
      Framework is in place for Python JSON debugging.
      Separate WATCH_NMEA from WATCH_RAW in the visible API.
      Accept $GN and $GL NMEA prefixes associated with GLONASS.
      Fix a glitch in dumping of device subtypes.
      Python client can now read VERSION and TPV objects from JSON output;
      hdop = sqrt(xdop**2 + ydop**2);
      Python client mode now handles DEVICE responses, and a documentation fix.
      Preliminary description of AIS message type 23.
      Fix slightly incorrect information on GLONASS talker IDs.
      Steps towards reading JSON skyviews.
      Python client code now reads skyview JSON correctly.
      Checkpoint efore considering incompatible changes.
      Clarify some AIS message 23 fields.
      Oops - cope well if there is no satellite data.
      Fully support AIS type 22 and 23 messages.  All regression tests pass.
      Typo fix.
      Handle AIS Type 21 Name Extension field.
      Update lists of known-good AIS decodings.
      Fix some source citations.
      New protocol can now ship profiling information, like old $.
      Document the TIMING response attributes properly.
      Update some description of driver internals.
      Introduce gps_send() to the C API.
      splint and typo cleanup.
      More splint cleanup.
      Get a name right, grrr.
      gps_query() is removed.
      Typo fix.  Codebase splints clean. All regression tests pass.
      Convert NMEA.txt to asciidoc markup and publish it as a web page.
      Corrections for AIS messages type 21 and 23.
      Re-enable JSON dumping.
      Document a limitation.
      AIS message type 21 now has a regression test.
      We now have regression tests for AIS Type 15 and 20 messages.
      Two more AIS regression tests, type 7 and 10.
      Another regression test for AIS message type 18.
      Another AIS type 8 regression test.
      Avoid error messages due to discontinuous lists.
      Ignore a generated file.
      gps_stream() gets a third argument for future expansion.
      Get rid of a trick macro that's not actually needed.
      Better description of new Python API.
      Correct an error in the libgps test main.
      Clean up the way the Python client code handles daemon socket closure.
      Improved Python client-mode documentation.
      C and Python libraries now speak both new and old protocol...
      Improvements to the logic for TIMINGS.
      gps_stream has a third argument now.
      Back out some of my second-to-last commit, it caused a regression.
      Pacify splint.
      Regression fixed.  All tests pass. Code splints clean.
      Refactoring step. All regression tests pass.
      Decorate TIMING objects with client-side timing information.
      Virtuous code removal  Old-protocol Z and $, never documented, are now 
      Revert most changes to since 21 Sep.
      Splice in the new stream method, that's backwards-compatible.
      Complete merge-forward of the iterator interface. All regression tests 
      Fix the documentation for the Python client.
      GPSD-NG code is no longer experimental.
      Verify that we can get JSON responses through the iterator interface.
      Fix some backwards-compatibility logic.
      More virtuous code removal.
      Make sure version member is always initialized.
      MND should be available at all firmware rev levels.
      Un-invert a test.
      Setting members of the driver state function during a wakeup event...
      splint and instrumentation cleanup.
      Uniform treatmen of VERSION response in poll() and next().
      Spaceplot function of gpsprof works with new protocol.
      Make sure we get a DEVICE message on assignment.
      Add a pointer to mrd's Perl library.
      Uninstrumented plot now works.
      Remove dead code.
      Profiling is no longer supported through the old interface.
      Get rid of some remnants of profiling code.
      Remove a flag associated with the old profiling feature.
      Beginning of revised profiling facility.
      Rip out most of the old profiling support.
      Raw plotting works.
      The raw plot works now.
      gpsprof is fully working with the new protocol.
      Remove some obsolete rate-limiting and clean up some report generation.
      Typo fix. All regression tests pass.
      Add some debugging machinery - we can now dump set flags in a convenient 
      Make this code a little more forgiving about the form of mask names.
      Ignore maskdump.c (generated) and typescript.
      Clean up the daemon mask values.  All regressions pass.
      Refactor the cycle-end tracking.  The AIS machinery should use it.
      Move climb computation from the NMEA driver to the error modeler.
      Introduced new LOG_DATA debugging level.
      Fix computation of epx and epy from eph.
      Adapt some drivers to use the new LOG_DATA level.  All regression tests 
      Remove a comment describing a dodgy calculation...
      Dead code removal.
      More DATA_LOG adaptations.  All regression tests pass.  Code splints 
      In the gpsdata structure, satellites -> satellites_visible
      Change some legends in printf strings...
      Have a single point of truth in the JSON skyview reports...
      Fix altitude interpretation in SiRF binary.
      Clean up after a name change.
      This patch fixes the end-of-test flakiness in the regression tests...
      Revert test/stable/zodiac.log.chk change from the last commit...
      Two more drivers now use the LOG_DATA convention. All regression tests 
      Zodiac driver adapted to do LOG_DATA.
      Clear up some driver-level confusion surrounding timestamp setting.
      More sensible mask dumping when the set is null.
      My bright idea to use timestamp() to set skyview times backfired...
      Fix a paste-o that made its way into many LOG_DATA format strings.
      The new instrumentation pays off.
      Condition out some dead code.
      Fix makefile typo that interfered with "make dist".
      Correct some history.
      Prevent the daemon from being confused by malformed sentences.
      Improve the output from the iTalk...
      Change some logging levels and reduce verbosity.
      Clean up the channel allocation logic.
      Fix an incorrect AIVDM fieldname.
      Correct a scaling factor in the Python decoder.
      Future schedule got messed up by the Berlios debacle.
      Avoid a core dump in the C++ library.  Suggestion by Don Weeks.
      Suppress unstable and negative timestamps in devices like the 
      Condition out some debugging stuff no longer used.
      Fix a corner case in the regress-driver filter.
      Address Berlios tracker bug #16299:
      Add COMMITTERS file.
      Ship so the build won't break.
      Document the no-close problem.
      In the wake of the change to use only C99 fixed-length types...
      Splint pacification, part 2.
      Packaging change requested by mrd;
      Silence some compiler warnings.  All regression tests pass.
      maskaudit goes to the main directory, since it's now part of the normal 
      Typo fix.
      Attempted fix for a recently-reported startup bug in the oncore driver.
      Typo fix.
      Revert Miroslav Lichvar's PPS patch of 27 Aug on Gary Miller's advice. the event_identified hook and fix a broken conditional.
      Address Berlios tracker bug #16408 ubx_driver: wrong packet length on 
      Fix driver bugs pointed out by Hakan Johannsen.
      Fix truly embarassing typo.
      Use booleans in cgps.c where appropriate. Move code so the
      Force -X11 inclusion in X_LIBS.
      Fix a silly off-by-one error.
      Prevent from trying to install manpages...
      Previous fix mysteriously didn't work.  Adding instrumentation.
      Allow quoted strings to pass the type check for character values...
      Add post_array state to the JSON parser, fixes an error in device-list 
      Dump the returned erro when ERR_SET is raised.
      Correct DEVICE invocations for gpsctl.c.
      mode -> native.
      Make some logging a bit more explicit.
      Enable option to condition out C++ support in libgps.c
      Merge HÃ¥kan Johansson's patch to allow ntpsh as non-root.
      HÃ¥kan Johansson writes:
      Rebuild regression tests to incorporate the embarassing-typo fix from 5 
      Splint cleanup.  All regression tests pass.
      Back out previous commit.
      Revert xgps to using old protocol while I chase a bug...
      Update START_SUBFRAME for latest IERS bulletin.
      Typo fix.
      Comment typo fix.
      More tweaks for making C++ support optional from edo1.
      vdop was mislabeled as zdop.
      Make it possible to pass a cycle change to gpsctl.
      Fixes for generated DEVICE calls.
      Improvements to core cycle-change logic, and MKT3301 support.
      Splint cleanup.
      Comment fixes.  All regression tests pass.
      Be a bit more generous about setting the observed mask...
      Avoid null-pointer issues.
      Add some instrimentation with which to detect JSON parser problems.
      Add a missing check field to a parser specification.
      More instrumentation for the JSON parseer.
      Comment fix.
      Compiler warning suppressions (the MKT3301->MTK3301 change wasn't 
      Avoid doing some invalid buffer copies that can cause really obscure 
      I found the bug in JSON parsing, so re-enable new protocol on all clients.
      Condition out the compass-mode code in cgps.
      cgps is converted to use new protocol.
      A start on an improved manual page.
      Whitespace fixes.
      Fix some whitespace glitches found while experimenting with indent(1).
      Give up on indent, documenting the reason.
      Fix some overealousnes code removal last night.
      Suppress a warning.
      cgps must collect device information from DEVICE notifications.
      Change WATCH so it aelicits a DEVICES in addition to the WATCH response.
      cgps now gets GPS type.
      Change 2-space indents to 4.
      More efficient use if vertical screenspace.
      Carry link to newest SiRF manual.
      Documentatyion and FAQ improvements.
      Anchor name fix.
      Markup fix.
      Add qualification on GPS-76 USB.
      Instrumentation for the client library.
      Keep validity flags for the client-side union in a valid state.
      Improved instrumentation.
      End pointer returned by the JSON parser needs to be passed back up...
      Consume whitespace following a JSON parse.
      Fix up some regression tests to track a recent API change.
      Fix an incorrect attempt to clear union state flags.
      Work around, and document, a subtle limitation of libgps.
      Typo fix.
      splint cleanup.
      Mac OS X cleanup passes from mledford.
      splint cleanup.
      Corrections for Garmin GPS-76b and an HTML bug in code generated by
      Documentation polishing.
      Add a protocol-transition comment.
      Recent TSIP changes require a rebuild of the trimble-lassen_iq-playacar 
      Resolve an OS X linkage problem.
      Beginning of workaround for OpenBSD socket problem.
      Comment fix.
      Comment fix.
      Document a dependency.
      Appease the malevolent autotools gods.
      Trying to be tricky about which things we hexdumped was a lose.
      Bump version to 2.90dev in preparation for release.
      Include the whole config file in gpsd.h.
      Arrange for a revision.h to be generated into release tarballs...
      Update the release instructions to include a check on revision.h.
      Make sure revision.h is in the tarball.
      Support for the Trinmble BD960/BX960
      Client library debugging can now nbe enabled at runtime.
      More client-debugging instrumentation.
      Remove incorrect comment.
      Mac OS port change - BUFSIZ, 8192 under Linux, is 1024 there.
      Undo a mistaken commit.
      Prevent Python clientlib from fruitlessly trying to evaluate lines of raw 
      Promise some structure stability.
      Require Python 2.6 (we need the json module).
      Fix documentation to match code.
      Fix an erroneous flags set that was breaking a regression test.
      The buffer_policy attribute of WATCH is gone.  Bring an example up to 
      Typo fix.
      Don't core-dump on (obsolete) M when no GPS is attached.
      Fix up a regression test of the sentence decoder.
      Add and document some client-side instrumentation.
      Pick up a dropped stitch near DBUS.
      Add batch mode capability to client library tester.
      Create a regression-test framework for the client library...
      Send components of a DEVICE response to the dev member...
      Record the DBUS prerequisites in INSTALL.
      Fix the DBUS build.
      Fix installation directions.
      Typo fix.
      Corrections to the AIS specification, and one fieldname.
      Last change to this (trying to asvoid parsing $ lines) hung regression 
      This generator file should have been in the repo long since.
      I missed the fact that is now generated from
      Deal with not having execute permission.
      Finish up the type -> ai_type name change in Message 21.
      Fix a test that was broken by the last fix to the json  client-side stuff.
      Note that libtool is broken.
      Document autotools breakage more exactly.
      Sigh, generate didn't have excute permissions either...
      Add missing string quote.
      python -> $(PYTHON)
      Forward-port fixed to make build work again under 2.64.
      Thread-hook support and documentation removed.
      Enable handling of multiple JSON objectts per line, and add a test for it.
      Add capability to skip on decoding errors.
      Add a verbosity option for debugging.
      Use -D uniformly for debug levels.
      Improved handling of message 23 in the Python decoder.
      Fix an argument-handling buglet.
      Document that a trailing @ should terminate a packed-six-bit AIS string.
      Make the Python decoder handle type 14 messages better.
      1.18 update of the decoding description.
      Add a note on sources.
      Added regression-test loads for AIS message types 12-14.
      There are client library tests now, so rename the 'stable' directory...
      Markup fix.
      Explain the leap-second time lag.
      Fix up gpsctl's code for setting serial parameters.
      Explain the SiRF timelag problem.
      Convert to new ASCIIDOC table syntax.
      Arrange for version and revision to be displayed by the -V options of
      More tweaks to make -V do the right thing.
      Add description of AIS Hub.
      ASCIIDOC polishing.
      Fix a minor glitch in the client-side debugging instrumentation.
      The -j option is no more, so gpspipe doesn't have to process or send it.
      Correct -V option and cosmetic fixes for gpspipe.
      Teach gpspipe to use new protocol.
      Oops, not everything needed got included in last commit.
      Refactor so the assembly of the gpspipe initi command is all done in one 
      Emitting well-formed JSON would sure help.
      Client side of gpspipe is done.
      Solve the libgps.0.0.0 problem.
      Looking for a file descriptor less than -1 is obviously wrong.
      Remove some debugging that gemiller probably doesn't need any more.
      gpspipe exposed an orthogonality failure.
      Fix a trivial bug in the gps_stream()/ entry point.
      Splint and compiler warning cleanup.
      Rip out the buggy sockfile usage in
      Fix so it will try to import simplejson if importing json fails.
      Introduce ENABLE_RARE flag for gps_STREAM(); gpspipe() will need it.
      Change the library private data member to void *...
      splint cleanup.
      Address Berlios tracker bug #16468:
      Typo fix.
      HTML metadata fix.
      Fix one of gdt's regression-test failures.
      Add "GPSD and Code Excellence" essay to white papers.
      Eliminate --disable-python.
      Remove some complexity.
      Require at least Python 2.4 for the build.
      This commit resolves the time overflows in the rgm3800 and mkt3301.
      Splint cleanup. All rests pass.
      Add a to-do item for 2.91.
      Add two more regression tests.  Pre-release doc cleanup.
      Make old protocol code interact correctly with the json policy flag.
      R command is fully working again.
      Ensure that revision.h is not installed.
      Version bump.  Time to ship 2.90.

Gary E. Miller (75):
      Fix SiRF Switch to NMEA, message ID 129, baud was in the wrong place.
      Output gps_send() failure messages.
      Fixed broken select() at startup.
      Improve logging a tad.
      Make class Devices parity type match what is really sent.
      Whoops, parity needs to default to 'N', ...
      I just discovered RFC 4627 which defines JSON.
      At least report when an error JSON is received.
      Add some gpsd side JSON error reporting.
      Add comment and remove redundant else.
      Fix typo in comment
      Fix gpsd_ppsmonitor() to handle very short (1uS) pulses.
      Add a comment so people understand how nice() works...
      Splint fix.
      Added some comments to explain the ntpd algorithm
      Fix test for nice() failure.  Found by Hakan Johansson.
      Whoops.  A bit too quick with the patch.
      Never give up totally looking for PPS,...
      Smll fixes and cleanups for PPS code.
      More ppsmonitor tweaks.
      Reorder the tests in pps_monitor...
      More refactoring of ppsmonitor().
      Fix #ifdef TSIP nesting from Chris Adams's TSIP-validation patch.
      Change logging in ntpshm_pps() so now all PPS logs are prefixed with PPS.
      Fix an oops pointed out by Hakan Johansson.
      Be a tad more liberal on accepting 5Hz pulese.
      Improve error messages for shmget() failures.
      More PPS cleanup
      Get current second from the last GPS timestamp.
      Shut up splint.  Per Hakan Johansson.
      Tweak a log message.
      Some packets do not end in \n, so make sure there is on in the log 
      Correct confusion of Garmin NMEA and Garmin Serial binary
      With this change I think you can nicely log all PPS/NTPD related logs 
this way:
      Fix last logging commit and add some more NTPD logging.
      More tagging of NTPD logs.
      Improve NTPD logging (and some code) in the ntpd parts of SiRF.
      Make fudge explicit in ntpshm_put(), and log it..
      Quite the ntpshm_alloc() logging a bit.
      Make the last patch compile.
      Time of arrival of SiRF binary messages jumps around too much.
      Ignore SiRF message type 0x04 for OOx purposes.
      Adjust Garmin Serial Binary NTPD fudge for 4800bps.
      Make 'gpsctl -b' work for SiRF binary.
      Only try to move to SiRF binary from NMEA if...
      Elegant solution to sirf_binary.mode_switcher() extern.
      Fix PPS precision at -16.
      Sorry, make the last patch compile.
      Improve comments and make dependency on /proc/bus/usb/devices optional.
      Further comment tweaks
      Fix setting of parity when changing SiRF from bianry to NMEA mode.
      Add SiRF NMEA manual to references
      Add SiRF NMEA doc Ver 2.2
      Chris Kuethe points out 8O1 is also valid.
      Add a SiRF comment
      Anotehr SiRF comment tweak.
      Fix the html for SiRF section in references.html
      A bit more NTP logging for SiRF
      Allow gpsctl -D to enable json debug.
      Make gpssctl debug output go to stderr.
      Enable json debugging with -D in gpsd
      Fix typo in last svn.
      stopbits is integer, not string.
      Tweak SiRF comment
      Log SiRF speed changes.
      Another logging statement, and a typo to fix speed setting.
      Add a warning message, 4800bps may be too slow for SiRF.
      Tweak the NMEA->ntpd time fudge for speed.
      More SiRF NMEA->ntpd fudge tweaks.
      When shutting  down the serial port keep the last baud rate...
      Fix gpsmon for glonass.  That is prefix GN instead of  GP.
      Remove untrue Garmin Speed comment.
      Prefix all SiRF driver messages with 'SiRF:" for easy logging.
      Further clenaup of SiRF comments.
      This patch fixes a segfault in the current trunk.

Greg Troxel (18):
      Don't try to include the non-existent m4 directory.
      cast times to unsigned long before printing with %lu.
      Generate and from templates...
      Replace multiline make rules that fail silently.
      Adjust type of local callback function to match type in the .h file.
      ignore generated tiles and
      Adjust prereq docs to the new reality.
      Distribute the '.in' versions of python scripts that generate sources...
      Add '#if 0' around code to exercise features removed from libgpsmm.h.
      These should not be marked executable, because they aren't actually 
      Cause 'make check' to actually run the regression tests.
      Substitute $PYTHON instead of assuming 'python' is python.
      point to hacking.html on website
      AM_PATH_PYTHON provided with automake can be too old.
      point out that the website bits are in the sources
      remove "$(srcdir)/" for files that belong in the build directory...
      Explicitly open gps.h in @srcdir@ instead of $CWD.
      Fix usage for VPATH builds, and hence make check.

Jeff Francis (1):
      Don't allow negative values when calculating magnetic headings.

Michael Ledford (4):
      Added logging in gpsd_error_model()...
      Increased logging precision gpsd_error_model().
      More logging to find regression test error on OS X. Getting closer.
      Removing logging put in place for debugging failing tests on Mac OS X.



reply via email to

[Prev in Thread] Current Thread [Next in Thread]