timers: consider implementation of timers which uses wait thread #6
Assignees
Comments
|
Assigning this to myself and thinking about it this week because of its relationship to real-time performance. this issue also mentions adding a timeout argument to |
jacquelinekay
added
in progress
This was referenced Jul 28, 2015
jacquelinekay
added
in review
dirk-thomas
added
in progress
jacquelinekay
added
in review
Closed
clalancette
added a commit
to clalancette/rclcpp
that referenced
this issue
Dec 8, 2021
* Remove template defaults that we never use. Signed-off-by: Chris Lalancette <clalancette@openrobotics.org> * Rename MessageT to PublishedType in the IntraProcessManager. It makes it easier to determine which type we are talking about. Signed-off-by: Chris Lalancette <clalancette@openrobotics.org> * Get all the types setup to push the conversion down to IPM. There really is no functional change here, just preparing to move the conversion down to the intra-process manager. Signed-off-by: Chris Lalancette <clalancette@openrobotics.org> * Move conversion down to the intra-process manager. Signed-off-by: Chris Lalancette <clalancette@openrobotics.org>
clalancette
added a commit
to clalancette/rclcpp
that referenced
this issue
Dec 21, 2021
* Remove template defaults that we never use. Signed-off-by: Chris Lalancette <clalancette@openrobotics.org> * Rename MessageT to PublishedType in the IntraProcessManager. It makes it easier to determine which type we are talking about. Signed-off-by: Chris Lalancette <clalancette@openrobotics.org> * Get all the types setup to push the conversion down to IPM. There really is no functional change here, just preparing to move the conversion down to the intra-process manager. Signed-off-by: Chris Lalancette <clalancette@openrobotics.org> * Move conversion down to the intra-process manager. Signed-off-by: Chris Lalancette <clalancette@openrobotics.org>
DensoADAS
pushed a commit
to DensoADAS/rclcpp
that referenced
this issue
Aug 5, 2022
* First implementation of writer * Extract storage interface and sqlite3 implementation * Add test for sqlite storage * Split main() and rosbag2::record() * Add close() method to Storage * Add getMessage() method and refactor test * Refactor SqliteStorage constructor and open() * Add linters * Fix uncrustify * Fix cpplint * Specify test working directory * Better error handling * Use gmock matchers for assertions * Add test fixture for SqliteStorage tests * Extract message retrieval in tests into separate method * Add integration test for rosbag2::record() * Add ignore files for empty packages * Introduce create() method and refactor open() * Use shared pointer of Storage instead of SqliteStorage * Remove getDatabaseHandle() method * Fix uncrustify * Improve storage interface and add storage factory * Remove need of sleep() from integration test by usage of std::future * Move deletion of test database from fixture constructor to destructor * Use sqlite3 directly in intergration test instead of own sqlite wrapper * Move rosbag2::record() into Rosbag2 class * Use the test name as database file name * Add build instructions to README * ros2GH-37 Rename camelCase methods to snake_case * Use common test fixture * Add RAII wrapper for sqlite API * Mock away sqlite from sqlite_storage test * Use more reasonable assert * Add test * Add virtual destructor to WritableStorage * Use file_name instead of database_name in StorageFactory * Implement saving of test files in a tmp directory for linux/Mac * Try to implement saving of test files in a tmp directory for Windows * Write and use proper gmock SqliteWrappe mock * Refactor integration test and get rid of promise/future where possible * Throw exception in resource aquisition constructors * Make SqliteWrapper destructor virtual * Refactor test fixture and update SqliteWrapper mock * Fix warning when moving a temporary object * ros2GH-38 Refactor integration test * ros2GH-38 Get rid of superfluous string constructor in emplace_back() * ros2GH-38 Assert also execute_query() argument in sqlite_storage_test * ros2GH-38 add StorageFactory test * ros2GH-38 Refactor rosbag2 Test Fixture * ros2GH-40 Add first implementation of a rosbag reader and publisher * ros2GH-40 Add StorageFactory test when reading non-existing file * ros2GH-40 Fix uncrustify * ros2GH-40 Minor cleanup of CMakeLists * ros2GH-40 Wrap sqlite statements * ros2GH-40 Remove superfluous import * ros2GH-40 Use better include * ros2GH-40 Add play integration test * ros2GH-40 Fix Warning when moving a temporary object in reading * ros2GH-40 Initialize database pointer to nullpointer * ros2GH-40 Fix reader integration test * ros2GH-40 Polish storage wrapper * Revert "ros2GH-40: Wrap sqlite statements" * ros2GH-38 Fix Test Fixture after rebase * ros2GH-38 Refactor read_integration_test and refix Windows conversion warning * ros2GH-38 Add StorageFactory test * Simplify storage factory test * ros2GH-38 Try to fix flaky test * ros2GH-38 Move rclcpp::shutdown() at the end * ros2GH-41 Fix windows warning due to virtual explicit operator bool * ros2GH-41 Use sqlite3 vendor package in rosbag2 * ros2GH-41 Stop linking tests to sqlite * ros2GH-41 Fix test fixture on Windows * ros2GH-41 Cleanup test fixture includes * ros2GH-41 Print test database name * ros2GH-41 Correctly determine temp dir on Windows * ros2GH-41 Show error message on sqlite_open failure * ros2GH-41 Actually create temp dir on Windows * ros2GH-41 Fix bool conversion warning in VS2015 build * Fix CMakeLists.txt after rebase * ros2GH-40 Implement workouround to fix flaky test * Update package.xml * Add gtest test dependencies to package.xml * ros2GH-40 Move to sqlite3_storage_plugin folder - The separation into the intended structure and plugin apis is not there yet. However, most code will stay in the storage plugin for sqlite3 file. - Proper separation of this code into storage plugin and rosbag layer will be done in bosch-io/aos-rosbag2#5. * ros2GH-40 Add TODO comments and small cleanup
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Currently timers are naively implemented using a thread per timer, but it should be possible to implement the timers using a timeout argument to the
ros_middleware_interface::waitfunction. This would potentially be much more efficient and at least more ellegant, as the total number of threads andros_middleware_interface::GuardConditionHandle's would be significantly reduced. A stepping stone to this, without requiring a timeout in theros_middleware_interface::waitfunction, would be to have one thread for all timers which does the same waiting mechanism, but using a sleep function not as a parameter toros_middleware_interface::wait.The text was updated successfully, but these errors were encountered: