ros / filters Goto Github PK

The change to how parameters are declared in ros2/rclcpp#1522 means there isn't a good way to declare a dynamic parameter that works on both foxy and galactic/rolling. Options I've considered:

1. Using SFINAE could to check for the existence of ParameterDescriptor::dynamic_typing. This type of code seems too unreadable.
2. In CMakeLists.txt, set a compiler definition for the version of rclcpp. Then use that to determine how to create a dynamic variable. This works within this package, but since the code is in a header file, anyone other package who uses it would need to define the same variable. Seems like it could cause problems downstream.

Since we should only really have bugfixes on foxy at this point, I think backporting from the ros2 branch shouldn't be too much of a burden. I'll create the new foxy branch and do a release from it (I assume that jenkins is using the releases to figure out what branch to test with which rosdistro....?)

Related to ANYbotics/grid_map#266

I am trying to build a library with multiple filter plugins, where in each filter filter_base.hpp is being included. I am getting linking errors (CI output) complaining about multiple definitions of filters::impl::normalize_param_prefix. Same thing is happening with filters::impl::load_chain_config.

I am getting this behaviour on ROS2 Rolling using bfb5917 and on Foxy using the binaries.

Exact errors:

/usr/bin/ld: CMakeFiles/grid_map_filters.dir/src/MinInRadiusFilter.cpp.o: in function `filters::impl::normalize_param_prefix(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >)':
MinInRadiusFilter.cpp:(.text+0xc0): multiple definition of `filters::impl::normalize_param_prefix(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >)'; CMakeFiles/grid_map_filters.dir/src/ThresholdFilter.cpp.o:ThresholdFilter.cpp:(.text+0xc0): first defined here

/usr/bin/ld: CMakeFiles/filters_demo.dir/src/FiltersDemo.cpp.o: in function `filters::impl::load_chain_config(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::shared_ptr<rclcpp::node_interfaces::NodeLoggingInterface> const&, std::shared_ptr<rclcpp::node_interfaces::NodeParametersInterface> const&, std::vector<filters::impl::FoundFilter, std::allocator<filters::impl::FoundFilter> >&)':
FiltersDemo.cpp:(.text+0xd00): multiple definition of `filters::impl::load_chain_config(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::shared_ptr<rclcpp::node_interfaces::NodeLoggingInterface> const&, std::shared_ptr<rclcpp::node_interfaces::NodeParametersInterface> const&, std::vector<filters::impl::FoundFilter, std::allocator<filters::impl::FoundFilter> >&)'; CMakeFiles/filters_demo.dir/src/filters_demo_node.cpp.o:filters_demo_node.cpp:(.text+0x210): first defined here

I think this could be fixed be making the functions mentioned above static as to make their scope local to their receptive files, does that sound right?

Please let me know if you need any further information, thanks in advance for your help :)

I have ported filters package for crystal. Request you to create new branch for it so that i can raise a Pull request.

As described by @v4n here, there are some pitfalls to the current filter finding strategy. Currently if there is no package name specified for a filter in the yaml, then the first filter with the specified string anywhere in its name is chosen:

That does output a warning, but given that it can create some really subtle errors when the wrong filter is chosen, it might be better if it actually errored out. For better or worse, many robots have lots of warnings output by various packages, and one more can be easily missed. I can think of a few possible solutions (these would probably only be applied in lunar and later ROS distros):

1. When no package is specified, require an exact string match (not just a substring) between the string specified in yaml and the name of the plugn (without the package)
2. Require specification of the full plugin name, including the package. Error out if the package name isn't there. This is the strictest option. I prefer it since it removes ambiguity from the start, but it could require a lot of people to update their robot configs.
3. Error out if more than one plugin matches the string from the yaml. In theory this would only break robot configurations that are already silently broken.

@tfoote @v4hn what do you guys think?

Using the deprecated pluginlib macros causes a warning to be printed for any node that directly or indirectly uses filters.

These should be updated to the new pluginlib API.

pluginlib WARNING: In file __FILE__ PLUGINLIB_DECLARE_CLASS is deprecated, please use PLUGINLIB_EXPORT_CLASS instead. You can run the script 'plugin_macro_update' provided with pluginlib in your package source folder to automatically and recursively update legacy macros.  Base = base_class_type, Derived = derived_class_type

Location:
/tmp/buildd/ros-hydro-filters-1.7.3-0precise-20130721-0336/src/transfer_function.cpp::34

XML definition forbids < inside strings and should be escaped by < Using Python I get errors parsing this file.

E.g: https://github.com/ros/filters/blob/hydro-devel/default_plugins.xml#L3

See: http://www.w3schools.com/xml/xml_syntax.asp

To reproduce:

$ rosinstall_generator desktop --rosdistro jade --deps --wet-only --upstream-development | grep filters
The following repositories with an unknown upstream will be ignored: filters, metapackages, rosbag_migration_rule

The warning message hints at unknown upstream.

Am I right or set_capacity() member is not implemented? I searched in the whole repo, only one result for this symbol, and it's the line at the link below.

https://github.com/ros/filters/blob/e33463aebca78471c0e52f45eaf31585b781af61/include/filters/realtime_circular_buffer.hpp#L83C8-L83C20

In this case, would you like me to provide a PR?

See the build status of #35 . CI could not install catkin :(

@tfoote do you mind if I do this?

I wasn't aware of this package until @gavanderhoorn pointed it out just now, but maybe this should live at https://github.com/ros-controls/ ? Thoughts @bmagyar?

@tfoote downstream packages need a release that includes #50

Is there a good reason for not having function

virtual bool update(T& data)=0;

in FilterBase?

It seems to me it might help getting the filter chain even more efficient.

Regarding backwards compatibility, it'd probably need to have a default implementation that calls the copying update(data_in, data_out), but filters that are capable of working in-place would get the chance to do so.

Could you please create ros2 branch so that we can use it to create pull request ?
We have just started filters porting to ROS2

why not have indigo-devel
how i use the filters in indigo

I am trying to build this using colcon build command. But it is not working, following error shows up:
`--- stderr: filters
CMake Error at CMakeLists.txt:26 (add_library):
add_library INTERFACE library requires no source arguments.

CMake Error at CMakeLists.txt:31 (target_link_libraries):
Cannot specify link libraries for target "realtime_circular_buffer" which
is not built by this project.

Failed <<< filters [6.26s, exited with code 1]`

It says INTERFACE library requires no source arguments. how can I resolve this without specifying the source files?

Currently getParam() is not overloaded for boolean type parameters. At the same time string parameters cannot mimic that behavior since the parser (?) recognize it as boolean and getParam() returns false as if parameter of requested type was not found.

@jonbinney did you mention you could release this? 😉

I believe the recent ros2/rclcpp#1522 breaks filters at runtime
When running FilterChain::configure I get:

scan_to_scan_filter_chain-1] terminate called after throwing an instance of 'rclcpp::exceptions::InvalidParameterTypeException'
[scan_to_scan_filter_chain-1]   what():  parameter 'filter1.name' has invalid type: cannot declare a statically typed parameter with an uninitialized value

There have been a few changes, most notably the addition of target-specific dependency info in cmake.

We are working on laser assembler porting to ROS2 but its dependent on Filters package.
So we have taken up this package for porting.

Hi everyone, I am trying to implement "Parameters On Set Parameter Callback" feature, which I can update the parameters through a callback function. However, it seems like if I use the GetParam method, the declared parameter is set to a read only state.

Is there a special reason why it is set to read only?

It is set at this line here.

If I set param through the terminal, I will get this output.

Setting parameter failed: parameter 'filter1.params.output_frame' cannot be set because it is read-only

I have install this under the src folder, When I use catkin build in this project, it always says no such file.

Hi,

I'm using Ubuntu and I can't find this package in apt search.

Almost all other packages I need to install can be find in apt search, this might be the only one that I need to build from source, so a apt package should be nice.

Hi all,

Trying to compile filters with ros2 humble and boost 1.71. Unfortunately, that is broken:

Starting >>> filters
--- stderr: filters
In file included from /home/dev_ws/src/filters/test/test_realtime_circular_buffer.cpp:35:
/home/dev_ws/src/filters/include/filters/realtime_circular_buffer.hpp: In instantiation of ‘filters::RealtimeCircularBuffer<T>::const_iterator filters::RealtimeCircularBuffer<T>::begin() const [with T = int; filters::RealtimeCircularBuffer<T>::const_iterator = boost::cb_details::iterator<boost::circular_buffer<int, std::allocator<int> >, boost::cb_details::const_traits<boost::cb_details::allocator_traits<std::allocator<int> > > >]’:
/home/dev_ws/src/filters/test/test_realtime_circular_buffer.cpp:64:25:   required from here
/home/dev_ws/src/filters/include/filters/realtime_circular_buffer.hpp:91:16: error: ‘const class boost::circular_buffer<int, std::allocator<int> >’ has no member named ‘cbegin’; did you mean ‘begin’?
   91 |     return cb_.cbegin();
      |            ~~~~^~~~~~
      |            begin
/home/dev_ws/src/filters/include/filters/realtime_circular_buffer.hpp: In instantiation of ‘filters::RealtimeCircularBuffer<T>::const_iterator filters::RealtimeCircularBuffer<T>::end() const [with T = int; filters::RealtimeCircularBuffer<T>::const_iterator = boost::cb_details::iterator<boost::circular_buffer<int, std::allocator<int> >, boost::cb_details::const_traits<boost::cb_details::allocator_traits<std::allocator<int> > > >]’:
/home/dev_ws/src/filters/test/test_realtime_circular_buffer.cpp:64:25:   required from here
/home/dev_ws/src/filters/include/filters/realtime_circular_buffer.hpp:100:16: error: ‘const class boost::circular_buffer<int, std::allocator<int> >’ has no member named ‘cend’; did you mean ‘end’?
  100 |     return cb_.cend();
      |            ~~~~^~~~
      |            end
make[2]: *** [CMakeFiles/realtime_buffer_test.dir/build.make:63: CMakeFiles/realtime_buffer_test.dir/test/test_realtime_circular_buffer.cpp.o] Error 1
make[1]: *** [CMakeFiles/Makefile2:569: CMakeFiles/realtime_buffer_test.dir/all] Error 2
make[1]: *** Waiting for unfinished jobs....
make: *** [Makefile:141: all] Error 2

This change was introduced in #65 . I confirmed that compilation with 5507bc5175cb74e825ee20b9dc19e67ef6a4a022 (the last commit before the merge of #65 is successful.

It looks like all of the dependencies for this are available in Melodic, so it would be great to get this released. Thanks in advance!