* Baseline test and force threads to yield.
* Add timer tracking for executor.
* Add locking and test happy-path exit conditions.
* Move logic to multi_threaded_executor
* Address reviewer feedback by reducing scope of set.
* Expand tolerance on testing.
* comment fixup
Otherwise it seemed to me like it would yield twice.
* Add tests for member callbacks
* Add tests for member callback in Test class (not working with gcc7)
* Uncomment test that was failing
* Linter fixup
* Split ParametrVariant into Parameter and ParameterValue
* Test expects ParameterTypeException
* get_parameter_value() -> get_value_message()
* Make to_parameter() const and rename to to_parameter_msg()
* Adding parameter array support
* PR feedback
* Matching changes in upstream branch
* EXPECT_EQ takes expected value a first argument and actual as second
* Revert "Revert "Store the subscriber, client, service and timer (#431)" (#448)"
This reverts commit 168d75cf1e.
* Convert all rcl_*_t types to shared pointers
Converts all rcl_*_t types in the memory allocation strategy to shared pointers to prevent crash happening when a subscriber is reset.
Issue: #349
* fixups
* Convert all rcl_*_t types to shared pointers
Converts all rcl_*_t types in the memory allocation strategy to shared pointers to prevent crash happening when a subscriber is reset.
Issue: #349
* fixup! Convert all rcl_*_t types to shared pointers
* fix { use on function definitions
We always put the { on a new line for function definitions and class declarations.
* Get logger name from rcl [direct]
* Get logger name from rcl [indirect]
* Update tests
* fixup on variable usage
* Move get_logger_name to NodeLogging interface
* Control shared/static linking via BUILD_SHARED_LIBS
* Remove rmw dependency
* Add for rclcpp_lifecycle too
* exec depend on ament_cmake is for normalize_path
* rclcpp logging still uses fprintf all over the place.
Remove all printf log lines and replace with RCLUTILS_LOG_XXX macros.
Issue: #438
* fixup include order
* change rcutils_time_point_value_t type from uint64_t to int64_t
* small style changes
* fix test time datatype
* Update time primatives to int64_t
* change time primitive datatype to signed
* A few more instances of UL to L
* Move clear wait set from after rcl_wait to ahead
Current code clear wait set after rcl_wait, it is not respond latency
friendly. In fact, clear wait set operation is not urgent, making sure it is
done before next rcl_wait should be fine.
Signed-off-by: jwang <jing.j.wang@intel.com>
* remove trailing whitespace
