1.13.2 (2020-06-08)

• fix shebang line for python3 (#212)
• Contributors: Mikael Arguedas

1.13.1 (2020-05-15)

• Fix ring45 test expectations (#211) Copying https://github.com/ros/geometry2/commit/04625380bdff3f3e9e860fc0e85f71674ddd1587
• import setup from setuptools instead of distutils-core (#209)
• Contributors: Alejandro Hernández Cordero, Shane Loretz

1.13.0 (2020-03-10)

1.12.1 (2020-03-10)

• Use process_time() for Python 3.8 compatibility (#205)
• Bump CMake version to avoid CMP0048 warning (#204)
• Add rostest include dirs (#195)
• Remove trailing semicolons from tf sources (#187) * [tf] Removed trailing semicolons after functions from all sources Used the -Wpedantic compiler flag to find all occurrences
• Allow to choose output precision in tf_echo (#186) * Allow to choose output precision in tf_echo
• update how c++11 requirement is added (#184)
• update install destination in CMakeLists.txt (#183) * export binary to right locations * specify archive and runtime destinations, update whitespace (#5)
• add visibility macro
• windows bring up, use ROS_DEPRECATED
• Remove signals from find_package(Boost COMPONENTS ...)
• fixing error of casting away constness in method void tfSwapScalarEndian(const tfScalar& sourceVal, tfScalar& destVal) in line 626 of Vector3.h (#179)
• Fix log output typo: message_notifier -> message_filter (#177) Almost all the log outputs use message_filter, except one. The warning text still referred to message_notifier. This commit fixes that.
• Contributors: C-NR, James Xu, Maarten de Vries, Martin Günther, Shane Loretz, Victor Lamoine, Yoshua Nava

1.12.0 (2018-05-02)

• Adapt to new xmlrpcpp header location (#164)

• Maintain & expose tf2 Buffer in shared_ptr for tf (#163) - Adds a tf2_ros::Buffer via a public accessor method to expose to customers of Transformer - Maintains the tf2_ros::Buffer in a shared_ptr to safely share access to the Buffer object - As this is targeting Melodic, adds c++11 compile flags to grant access to std::shared_ptr's - Reorders the include_directories in the CMakeLists to ensure the headers exposed in this package are read before the system installed catkin_INCLUDE_DIRS (otherwise changes to tf source headers are never detected during a catkin_make on a system with ros-*-geometry installed)

  System Message: WARNING/2 (<string>, line 46); backlink

  Inline emphasis start-string without end-string.

• Prevent rates that result in core dump (0.0) or no limit on update rate at all (<0.0) #159 (#160)

• Fix empty yaml parsing (#153) Fixes #152 The empty yaml was coming through as a list not a dict so was breaking the expectations. I used the shorthand or {} since I know any valid data won't evaluate to zero. A more complete solution is described here: https://stackoverflow.com/a/35777649/604099

• Make python scripts Python3 compatible. (#151) * Python 3 fixes * Prefer str.format over operator % and small python3 fixes.

• Contributors: Ian McMahon, Lucas Walter, Maarten de Vries, Tully Foote

1.11.9 (2017-07-14)

• Replace legacy python code with appropriate calls to tf2_ros (#149) (#134)
• Replace deprecated Eigen module with Eigen3
• Update minimum version for run dependency on tf2_ros
• Add support for static_transforms in tf_monitor. Fixes #136 with #134 for tf_echo and view_frames.
• Pass through allFramesAsDot time argument optionally.
• remove vestigial includes. Fixes #146 (#147)
• Commented code caused error in documentation (#142)
• [doc] Add migration notice in manifest. (#129)
• Fix "stdlib.h: No such file or directory" errors in GCC-6
• Fix error for null conversion.
• Change version regex for graphviz in view_frames
• fix for issue in getAngleShortestPath(), closes #102
• Contributors: AndyZe, Edward Venator, Hodorgasm, Isaac I.Y. Saito, Michael Korn, Mike Purvis, Tom Moore, Tully Foote, Timo Röhling

1.11.8 (2016-03-04)

• Update assertQuaternionValid to check for NaNs
• Remove outdated manifest loading in python files
• update unit tests to catch https://github.com/ros/geometry_experimental/issues/102
• Contributors: Chris Mansley, Michael Hwang, Tully Foote

1.11.7 (2015-04-21)

• add a unit test for pytf wait_for_transform
• removed msg serv installation from cmakelists
• generated autodoc
• Fixed Vector3 documentation
• display RPY in both radian and degree
• Fixed command line arguments
• using TimeStamp and FrameId in message filter this allows to use tf::MessageFilter with pcl::PointCloud<T> see #55
• Added and optional third argument to specify publishing frequency
• Contributors: Adnan Munawar, Brice Rebsamen, Jackie Kay, Tully Foote, Ying Lu

1.11.6 (2015-03-25)

• reenable python tests
• Broadcaster: Rospy fix #84. Add sendTransformMessage.
• Contributors: Tully Foote, lsouchet

1.11.5 (2015-03-17)

• Strip leading slash get parent #79
• Make frameExists strip leading slash going into tf2.`#63 <https://github.com/ros/geometry/issues/63>`_
• Update broadcaster.py, Added ability to use TransformStamped
• update view_frames to use AllFramesAsDot(rospy.time.now()) #77
• Contributors: David Lu!!, Gaël Ecorchard, Kei Okada, Tully Foote

1.11.4 (2014-12-23)

• Install static lib and remove test for Android
• Larger default queue size in broadcaster With queue_size=1 when two transforms are sent in quick succession, the second is often lost. The C++ code uses a default queue_size of 100, so switch to that default here as well. If that is not appropriate, a queue_size constructor argument is provided.
• Update package.xml
• add rate parameter to tf_echo
• Added check for normalized quaternion in roswtf plugin
• Contributors: David Lu!!, Gary Servin, Kevin Hallenbeck, Stephan Wirth, contradict

1.11.3 (2014-05-07)

• convert to boost signals2 following ros/ros_comm#267 Fixes #23. Requires ros/geometry_experimental#61 as well.
• add rospy publisher queue_size argument ros/ros_comm#169
• add queue_size to tf publisher ros/ros_comm#169
• make rostest in CMakeLists optional (ros/rosdistro#3010)
• Contributors: Lukas Bulwahn, Tully Foote

1.11.2 (2014-02-25)

• fixing test linking
• Contributors: Tully Foote

1.11.1 (2014-02-23)

1.11.0 (2014-02-14)

• TF uses ros::MessageEvent to get connection information
• Contributors: Kevin Watts, Tully Foote

1.10.8 (2014-02-01)

• Port groovy-devel patch to hydro-devel
• Added rosconsole as catkin dependency for catkin_package
• Add rosconsole as runtime dependency
• Contributors: Michael Ferguson, Mirza Shah

1.10.7 (2013-12-27)

• fix bug in tf::Matrix3x3::getEulerYPR() Fixes a bug in tf::Matrix3x3::getEulerYPR() implementation's handling of gimbal lock cases (when the new x axis aligns with the old +/-z axis).
• add test that demonstrated bug in tf::Matrix3x3 tf::Matrix3x3::getEulerYPR() has a bug which returns an incorrect rpy for certain corner case inputs. This test demonstrates that bug.
• Fix const correctness of tf::Vector3 rotate() method The method does not modify the class thus should be const. This has already been fixed in Bullet itself.
• add automatic tf buffer cleaning on bag loop for python This logic was already implemented for c++ but not for the python module.
• Contributors: Acorn Pooley, Timo Rohling, Tully Foote, v4hn

1.10.6 (2013-08-28)

• switching to wrapper scripts which will provide a deprecation warning for #3
• add missing roswtf dependency to really export the plugin (fix #27)
• Update listener.py Fix the tf listener service exception in rospy. See: http://answers.ros.org/question/10777/service-exception-using-tf-listener-in-rospy/
• Fix MessageFilter race condition If MessageFilter does not explicitly stop its callback timer when it's being destroyed, there is a race condition when that timer is processed in a callback queue run by a different thread. Specifically, maxRateTimerCallback() may be called after messages_mutex_ has been destroyed, causing a unrecoverable error.

1.10.5 (2013-07-19)

• tf: export dependency on tf2_ros Fixes #21
• added run dependency on graphviz closes #19

1.10.4 (2013-07-11)

• fixing erase syntax
• resolving https://github.com/ros/geometry/issues/18 using implementation added in tf2::BufferCore, adding dependency on next version of tf2 for this

1.10.3 (2013-07-09)

• fixing unittest for new resolve syntax

1.10.2 (2013-07-09)

• strip leading slashes in resolve, and also any time a method is passed from tf to tf2 assert the leading slash is stripped as well. tf::resolve with two arguments will end up with foo/bar instead of /foo/bar. Fixes https://github.com/ros/geometry_experimental/issues/12
• added two whitespaces to make message_filter compile with c++11 more on this here: http://stackoverflow.com/questions/10329942/error-unable-to-find-string-literal-operator-slashes
• using CATKIN_ENABLE_TESTING to optionally configure tests in tf

1.10.1 (2013-07-05)

• updating dependency requirement to tf2_ros 0.4.3
• removing unused functions removing unused private methods removing max_extrapolation_distance_ removing unused data storage _frameIDs frameIDS_reverse frame_authority_ removing cache_time from tf, passing through method to tf2 buffer_core removing unused variables frames_ and frame_mutex_ and interpolating_ removing unused mutex and transformchanged signaling commenting on deprecation of MAX_EXTRAPOLATION_DISTANCE

1.10.0 (2013-07-05)

• adding versioned dependency on recent geometry_experimental changes
• fixing test dependencies
• fixing callbacks for message filters
• remove extra invalid comment
• dedicated thread logic all implemented
• removing commented out code
• mostly completed conversion of tf::TransformListener to use tf2 under the hood
• lookuptwist working
• tf::Transformer converted to use tf2::Buffer under the hood. passing tf_unittest.cpp
• making tf exceptions typedefs of tf2 exceptions for compatability
• first stage of converting Transformer to Buffer
• switching to use tf2's TransformBroadcaster
• adding dependency on tf2_ros to start moving over contents
• fixing unit tests

1.9.31 (2013-04-18 18:16)

1.9.30 (2013-04-18 16:26)

• Adding correct install targets for tf scripts
• Removing scripts from setup.py install

1.9.29 (2013-01-13)

• use CATKIN_DEVEL_PREFIX instead of obsolete CATKIN_BUILD_PREFIX

1.9.28 (2013-01-02)

1.9.27 (2012-12-21)

• set addditional python version
• added license headers to various files

1.9.26 (2012-12-14)

• add missing dep to catkin

1.9.25 (2012-12-13)

• add missing downstream depend
• update setup.py

1.9.24 (2012-12-11)

• Version 1.9.24

1.9.23 (2012-11-22)

• Releaseing version 1.9.23
• tf depended on angles but did not find_package it

1.9.22 (2012-11-04 09:14)

1.9.21 (2012-11-04 01:19)

1.9.20 (2012-11-02)

1.9.19 (2012-10-31)

• fix catkin function order
• Removed deprecated 'brief' attribute from <description> tags.

1.9.18 (2012-10-16)

• tf: Fixed wrong install directory for python message files.
• tf: fixed bug where generated python message code was not being installed.
• tf: added setup.py file and changed CMakeLists.txt to install python files and bound library (_tf.so, also known as pytf_py in CMakeLists.txt) which must have been missed during the previous catkin-ization.

1.9.17 (2012-10-02)

• fix several dependency issues

1.9.16 (2012-09-29)

• adding geometry metapackage and updating to 1.9.16

1.9.15 (2012-09-30)

• fix a few dependency/catkin problems
• remove old API files
• comply to the new catkin API

1.9.14 (2012-09-18)

• patch from Tom Ruehr from tf sig
• patch from #5401 for c++0x support

1.9.13 (2012-09-17)

• update manifests

1.9.12 (2012-09-16)

• use the proper angles package

1.9.11 (2012-09-14 22:49)

• no need for angles anymore

1.9.10 (2012-09-14 22:30)

• no need for bullet anymore

1.9.9 (2012-09-11)

• update depends
• minor patches for new build system

1.9.8 (2012-09-03)

• fixes for groovy's catkin

1.9.7 (2012-08-10 12:19)

1.9.6 (2012-08-02 19:59)

• changing how we install bins

1.9.5 (2012-08-02 19:48)

• fix the header to be compiled properly
• using PROGRAMS insteas of TARGETS

1.9.4 (2012-08-02 18:29)

1.9.3 (2012-08-02 18:28)

• forgot to install some things
• also using DEPENDS

1.9.2 (2012-08-01 21:05)

• make sure the tf target depends on the messages (and clean some include_directories too)

1.9.1 (2012-08-01 19:16)

• install manifest.xml

1.9.0 (2012-08-01 18:52)

• catkin build system
• remove bullet dep
• fix bug #5089
• add link flag for OSX
• tf: MessageFilter: added public getter/setter for queue_size_
• adding btQuaternion constructor for ease of use
• fixing method naming for camelCase and adding bt* Constructor methods
• tf.tfwtf now uses rosgraph.Master instead of roslib
• Added tf and angles to catkin
• cleanup up last errors
• SIMD_ -> TFSIMD_ defines to not conflict
• write in bullet assignment and return methods
• executable bit on conversion script
• changing defines from BT to TF
• removing BULLET_VERSION info
• changing all bt* to tf* in LinearMath to avoid collisions
• convert btScalar to tfScalar to avoid definition conflicts
• deleting GEN_clamp and GEN_clamped as they're unused and would conflict
• non conflicting minmax functions
• the migration script
• applied bullet_migration_sed.py to LinearMath include dir with namespaced rules and everything with Namespaced rules and all 152 tests pass
• removing all BT_USE_DOUBLE_PRECISION ifs and hardcoding them to the double case
• adding tf namespaces to moved files
• breaking bullet dependency
• removing redundant typedefs with new datatypes
• moving filenames to not collide in search and replaces
• changing include guards
• moving linear math into tf namespace
• copying in bullet datatypes
• switching to a recursive mutex and actually holding locks for the right amount of time. ticket:5
• Giving error message when time cache is empty for lookup failures
• Moving lct_cache_ to local variable from class member. As class member, using this variable makes lookupTransform not thread-safe
• velocity test precision a little lower requirements
• Fix to error message for earliest extrapolation time exception, ros-pkg5085
• Fixing epsilon to prevent test failures
• Reducing epsilon value for velocity tests
• add missing empty_listener.cpp file
• Not calling ros::Time::now() in tf.cpp, causes problems with pytf
• fix for ROS_BREAK include
• Adding faster point cloud transform, as specified in ros-pkg`#4958 <https://github.com/ros/geometry/issues/4958>`_
• Cache unittest reenabled
• Adding speed_test from tf2 to check lookupTransform/canTransform
• Josh's optimizations from tf2 merged into tf. Tests pass
• Benchmark test includes tests of lookupTransform
• Adding ros::Time::init to benchmark test
• Testing compound transforms with lookupTransform
• Adding helix test of time-varying transforms, with interpolation, to test lookupTransform
• Moving test executables to bin/. Cleanup in tf_unittest. Removed deprecated calls to bullet, added 'ring45' test from tf2 as lookupTransform test
• patch for #4952
• kevin's patch for #ros-pkg4882
• Fix for TransformListener hanging on simulation shutdown, #4882
• removing old srv export
• removing old srv includ path
• this should never have been passing in an error string here -- likely one of the reasons MessageFilter is so slow
• Adding to author list to create branch
• removing reset_time topic and catching negative time change to reset the tf buffer
• #4277 transformPointCloud
• revert patch that uses ros::ok in waitForTransform. ticket #4235
• make tf unittest a ros node
• fix lockup in waitForTransform. ticket 4235
• reverting r30406 and r30407, they are redundant with standardized functionality and break previous functionality
• sse detection #4114
• tf: change_notifier should sleep after an exception
• created common place for ROS Pose/Point/Quaternion to numpy arrays transformations
• added TransformBroadcaster.sendTransform for PoseStamped
• one more patch for #4183
• new unit test
• waitforTransform now polls on walltime to avoid ros::Time initialization issues. basic unit test for waitForTransform in python.
• fix for stricter time
• fix ros::Time unit test problem with ROS 1.1.9
• #4103 method getTFPrefix() added, documented, tested
• moving patch to trunk from tag r30172
• Added Ubuntu platform tags
• Update MessageFilter to use traits and MessageEvent
• #4039, moved PoseMath from tf to tf_conversions
• #4031 add lookupTwist and lookupTwistFull
• fixing zero time edge case of lookupTwist, thanks james
• commenting debug statement
• Typo in comment
• documentation
• fixing up unit tests
• lookup twist for #4010
• commenting twist test while the code is being refactored
• removing transform twist as per api review in ticket #4010
• Added doctest for PoseMath creation from message
• Doc for PoseMath
• Double module tf
• Remove expect_exception
• comment for operator
• opeartor == for StampedTransform too #3990
• First cut at posemath
• adding operator== to Stamped<T> with unit tests
• adding methods for vectorized publishing of transforms #3954
• fix thread-safety of add()
• Re-add message filter test that was accidentally removed when the message notifier was deleted
• Fix message filter in the case where messages are arriving faster than the update timer is running (exacerbated by rosbag play --clock not actually broadcasting the clock at 100hz). (#3810)
• Tiny refactor for callerid->authority
• #3942 testcase
• Add doc for Transformer.clear
• Missing initializer from TransformListener
• New test test_cache_time
• fixing quaternion checking and adding unittests #3758
• review status #3776
• tf: change_notifier now supports multiple frames; publishes tfMessages
• passing basic tests for transformtwist
• adding transformTwist method
• all tests passing on lookupVelocity
• tests for values calculated by hand
• linear velocity to multiple other targets
• expanding to all three dimentions and asserting others are zero
• first cut velocity, basic test architecture layed out.
• searchparam when publishing
• noting deprecations better and changing frame_id to frame_name for unresolved
• removing /tf_message since it's been deprecated
• returning remap capability to remap #3602
• inlining helper function
• tf: changed manifest to have lower-case tf
• comment
• more documentation
• adding helper function for getting tf_prefix
• patches for tf_monitor to correctly display the chain, thanks for the help Sachin.
• asserting that incoming frameids are resolved, currently at debug level as this is not fully implemented in othe code. This level will escalate slowly as compliance is increased #3169
• not using my own deprecated function
• more usage
• tf_echo usage expanded
• fixing typo in documentation
• removing include of message_notifier
• removing deprecated message_notifier #3046
• removing deprecated data type and constructor #3046
• removing deprecated sendTransform calls
• fixing test for usage of deprecated APIs #3046
• removing deprecated setTransform method #3046
• removing deprecated lookupTransform methods #3046
• removed deprecated canTransform method #3046
• removing deprecated canTransform #3046
• removing deprecated transform_sender #3046
• removing deprecated transformStampedMsgToTF and transformStampedTFToMsg #3046
• fixing startup race condition #3168
• adding InvalidArgument exception for transformMethods, currently it only throws if w in quaternions are w <= 0 or w > 1 #3236
• reving for release
• commenting all velocity work for it's not ready to be released
• adding in deprecated call which I removed accidentally
• renaming tf::remap to tf::resolve as per #3190 with backwards compatability. Also Standardizing to only do searchparam at startup #3167
• Switch MessageFilter back to using a Timer instead of a WallTimer, since the time-jumping is now fixed (#2430)
• adding createQuaternionFromRPY method to help deprecation transition #2992
• Added specific tes for quaternion types
• Switching refernece frame and moving frame ordering for lookup transform call to actually be correct
• adding test to the back
• fixing lookupVelocity special cases for zero time
• documention improvements
• Doc clarifications
• removing debugging
• lookupVelocity Python first cut
• transformVector3
• switching tf_prefix to searchParam so you can set it for a whole namespace #2921
• removing .py extension from script
• simpler topic name
• adding tf_remapping script to remap frame ids #870
• fixing manifest loading to right package
• uncommenting lookup velocity and fixing implementation
• removing redundant angles package dependency #3334
• Patch from #3337
• fixing ~ usage
• commenting out lookupvelocity while it's still not working for release of patches
• angles needed for velocity lookup
• Switch from to_seconds to to_sec, #3324
• updating for 0.10 changes to python and hudson
• fixing deprecated to_seconds call in tfwtf
• merging 0.4x changes into trunk
• a first trial of lookupVelocity
• added createQuaternionMsgFromRollPitchYaw helper function
• removing wait_for_transform_death test from default, for it doesn't work under release
• switching to Release from Debug
• fixing usage message of static_transform_sender
• Warn about received messages with frame_ids that are not fully qualified, but resolve them locally
• moving deprecation note to top of summary
• • Remap target frames (#3119)
  • Throw out messages immediately if they have an empty frame_id
• fixing display of chain to show all links
• documentation for #2072
• fixing frequency output of tf_monitor
• making remapping on send more consistent
• removing unused variable
• Doxygen comments for the failure reasons
• Add a failure callback to tf::MessageFilter
• fixing #2990 deprecated ~ call
• update tf error strings. Still need review and user testing
• notifier should subscribe to tf and tf_message
• doc: updated setTransform to properly list child_frame_id
• Doc clearer on exceptions
• restoring caller_id to graph view in python
• Set daemon on listener thread
• better command line outputs
• Removed turtlesim reference from python broadcaster
• removing useages of deprecated bullet APIs
• Add rosdoc to manifest
• Fix build break
• New Sphinx docs
• changing display of legend to be above the tree
• make output consistent with view frames
• tweak output of tf_echo
• tweek output of tf_echo
• update output string
• update output of view frames
• make tf_echo wait for up to one second before throwing exceptions
• Fixes for pytf: exception distinction, waitForTransform, threaded listener
• Switch MessageFilter back to a WallTimer... shouldn't have been checked in with my last checkin
• Remove last remnants of Node use
• Fix compiler warnings
• removing last warnings relating to #2477
• tf monitor working, and a little bit cleaner display
• fixing useage of ~ params
• cleaning up tf_echo output
• fixing warning
• static_transform_publsher replacing transform_sender for backwards compatability, and fixing new StampedTransform
• update tf description
• remove extra / in method def. Ticket #2778
• fixed deprecation of Stamped<> 4 constructor vs 3 constructor. and switched three usages #2477
• converting transformPointCloud to use new StampedTransform #2477
• fixing warnings related to #2477
• internally switching to StampedTransform for #2477 cleanup
• fixing usage of Stamped<Transform> to StampedTransform
• switching Stamped<btTransform> to StampedTransform, deprecating usage, and changing all APIs to the new one with backwards compatabilty #2477. It's working but lots of warnings left to fix
• removing warning
• fixing deprecated function call usage
• one less node API call
• one less node usage
• fixing urls for new server
• Rename tf message from tf_message to tf. Listener is backwards compatible, broadcaster is not. See ticket #2381
• migration part 1