config_utilities

config_utilities is a minimal but powerful C++ library, providing tools to parse, verify, and print C++ config structs and configurable objects.

Table of contents

• Credits
• Why config_utilities?
• Installation
• How to config_utilities?
• Example Projects using config_utilities

Credits

This library was developed by Lukas Schmid and Nathan Hughes at the MIT-SPARK Lab, based on functionalities in ethz-asl/config_utilities and Hydra, and is released under a BSD-3-Clause License! Additional contributions welcome! This work was supported in part by the Swiss National Science Foundation and Amazon.

Why config_utilities?

Among many other, the key features of config_utilities include:

• Minimal dependencies: Only C++17 standard library and yaml-cpp.

• Declare any struct a config, also from external projects:

  namespace external_project {
      void declare_config(ExternalObject& config); // that's all!
  }   // namespace external_project

• Minimal and clear, human readable interfaces for config definitions:

  void declare_config(MyConfig& config) {
      using namespace config;
      name("MyConfig");                             // Name for printing.
      field(config.distance, "distance", "m");      // Field definition.
      check(config.distance, GT, 0.0, "distance");  // Ensure 'distance > 0'.
  }

• Everything related to a config is defined in a single place.

  void declare_config(MyConfig& config) { /* ALL the information about a config is here */ }

• Parse any declared config from various data sources, without pulling in dependencies on data sources into core libraries:

  core.cpp {
      struct MyConfig { ... };
      void declare_config(MyConfig& config) { ... };
  }  // Depends only on C++ and yaml-cpp.
  
  application.cpp {
      MyConfig config1 = fromYamlFile<MyConfig>(file_path);
      MyConfig config2 = fromRos<MyConfig>(node_handle);
  }  // Pull in dependencies as needed.

• Automatically implements frequently used functionalities, such as checking for valid parameters and formatting to string:

  MyObject(const MyConfig& config) : config_(checkValid(config)) {
      // Do magic with config_, which has valid parameters only.
      std::cout << config_ << std::endl;
  }

• Automatic run-time module creation based on specified configurations:

  static auto registration = Registration<Base, MyDerived>("my_derived");
  
  const std::string module_type = "my_derived";
  std::unique_ptr<Base> module = create(module_type);

• Informative, clear, and human readable printing of potentially complicated config structs:

  =================================== MyConfig ===================================
  distance [m]:                 -0.9876
  A ridiculously long field name that will not be wrapped: Short Value (default)
  A ridiculously long field name that will also not be wrapped:
                                A really really really ridiculously long string th
                                at will be wrapped. (default)
  A really really really really really really ridiculously long field name that wi
  ll be wrapped:                A really really really ridiculously long string th
                                at will also be wrapped. (default)
  vec:                          [5, 4, 3, 2, 1]
  map:                          {b: 2, c: 3, d: 4, e: 5}
  mat:                          [[1, 2, 3],
                                 [4, 5, 6],
                                 [7, 8, 9]]
  my_enum:                      B
  sub_config [SubConfig] (default):
     f:                         0.123 (default)
     s:                         test (default)
     sub_sub_config [SubSubConfig] (default):
        color:                  [255, 127, 0] (default)
        size:                   5 (default)
  ================================================================================
  

• Verbose warnings and errors if desired for clear and easy development and use:

  =================================== MyConfig ===================================
  Warning: Check failed for 'distance': param > 0 (is: '-1').
  Warning: Check failed for 'subconfig.f': param within [0, 100) (is: '123').
  Warning: Failed to parse param 'vec': Data is not a sequence.
  Warning: Failed to parse param 'mat': Incompatible Matrix dimensions: Requested
          3x3 but got 3x4.
  Warning: Failed to parse param 'my_enum': Name 'D' is out of bounds for enum wit
          h names ['A', 'B', 'C'].
  Warning: Failed to parse param 'uint': Value '-1' underflows storage min of '0'.
  ================================================================================
  
  [ERROR] No module of type 'NotRegistered' registered to the factory for BaseT='d
  emo::Base' and ConstructorArguments={'int'}. Registered are: 'DerivedB', 'Derive
  dA'.
  
  [ERROR] Cannot create a module of type 'DerivedA': No modules registered to the
  factory for BaseT='demo::Base' and ConstructorArguments={'int', 'float'}. Regist
  er modules using a static config::Registration<BaseT, DerivedT, ConstructorArgum
  ents...> struct.
  

Installation

This package is compatible with catkin and catkin_simple. Just clone it into your workspace and you should be all set!

cd ~/catkin_ws/src
git clone git@github.com:MIT-SPARK/config_utilities.git
catkin build config_utilities

If you want to build and install without catkin/ROS, that is easy, too! Just clone this repository and build via CMake:

git clone git@github.com:MIT-SPARK/config_utilities.git
cd config_utilities/config_utilities
mkdir build
cd build
cmake ..
make -j

# optionally install this package
sudo make install

How to config_utilities

We provide detailed introductions about everything you need to know about config_utilities in the following tutorials and some verbose example demos that you can run.

The (non-ros) demos can be run via the run_demo.py utility in the scripts directory. If you are building this library via catkin, you can run one of the following to see the results of one of the corresponding demo files:

python3 scripts/run_demo.py config
python3 scripts/run_demo.py inheritance
python3 scripts/run_demo.py factory

ℹ️ Note
If you're building via cmake, you can point run_demo.py to the build directory with -b/--build_path.

The ros demo can be run via:

roslaunch config_utilities demo_ros.launch

If you are looking for a specific use case that is not in the tutorials or demos, chances are you can find a good example in the tests/ directory!

Example Projects using config_utilities

Many cool projects are already using config_utilities! If you are using config_utilities and would like to be featured, please open a pull request or reach out!

Academic Open-source Projects

These projects are openly available, check them out to see what config_utilities can do!

• MIT-SPARK/Khronos
• MIT-SPARK/Hydra

Use in Industry

Several leading companies and start-ups in robotics have let us know that they use config_utilities, but we cannot currently list them.