LogicEngine

class rlogic::LogicEngine

Central object which creates and manages the lifecycle and execution of scripts, bindings, and all other objects supported by the Ramses Logic library. All objects created by this class’ methods must be destroyed with destroy!

  • Use the create[Type] methods to create various objects, use destroy() to delete them.

  • Use link and unlink to connect data properties between these objects

  • use update() to trigger the execution of all objects

Public Functions

LogicEngine() noexcept

Constructor of LogicEngine.

~LogicEngine() noexcept

Destructor of LogicEngine

template<typename T>
Collection<T> getCollection() const

Returns an iterable rlogic::Collection of all instances of T created by this LogicEngine. T must be a concrete logic object type (e.g. rlogic::LuaScript) or rlogic::LogicObject which will retrieve all logic objects created with this LogicEngine (see rlogic::LogicObject::as<Type> to convert to concrete type).

Return

an iterable rlogic::Collection with all instances of T created by this LogicEngine

template<typename T>
const T *findByName(std::string_view name) const

Returns a pointer to the first occurrence of an object with a given name of the type T. T must be a concrete logic object type (e.g. rlogic::LuaScript) or rlogic::LogicObject which will search any object with given name regardless of its type (see rlogic::LogicObject::as<Type> to convert to concrete type).

Return

a pointer to the logic object, or nullptr if none was found

Parameters

  • name: the name of the logic object to search for

template<typename T>
T *findByName(std::string_view name)

Returns a pointer to the first occurrence of an object with a given name of the type T. T must be a concrete logic object type (e.g. rlogic::LuaScript) or rlogic::LogicObject which will search any object with given name regardless of its type (see rlogic::LogicObject::as<Type> to convert to concrete type).

Return

a pointer to the logic object, or nullptr if none was found

Parameters

  • name: the name of the logic object to search for

const LogicObject *findLogicObjectById(uint64_t id) const

Returns a pointer to the first occurrence of an object with a given id regardless of its type. To convert the object to a concrete type (e.g. LuaScript) use rlogic::LogicObject::as<Type>() e.g.: auto myLuaScript = logicEngine.findLogicObjectById(1u)->as<LuaScript>()); Be aware that this function behaves as dynamic_cast and will return nullptr (without error) if given type doesn’t match the objects type. This can later lead to crash if ignored.

Return

a pointer to the logic object, or nullptr if none was found

Parameters

  • id: the id of the logic object to search for

LogicObject *findLogicObjectById(uint64_t id)

Returns a pointer to the first occurrence of an object with a given id regardless of its type. To convert the object to a concrete type (e.g. LuaScript) use rlogic::LogicObject::as<Type>() e.g.: auto myLuaScript = logicEngine.findLogicObjectById(1u)->as<LuaScript>()); Be aware that this function behaves as dynamic_cast and will return nullptr (without error) if given type doesn’t match the objects type. This can later lead to crash if ignored.

Return

a pointer to the logic object, or nullptr if none was found

Parameters

  • id: the id of the logic object to search for

LuaScript *createLuaScript(std::string_view source, const LuaConfig &config = {}, std::string_view scriptName = "")

Creates a new Lua script from a source string. Refer to the rlogic::LuaScript class for requirements which Lua scripts must fulfill in order to be added to the LogicEngine. You can optionally provide Lua module dependencies via the config, they will be accessible under their configured alias name for use by the script. The provided module dependencies must exactly match the declared dependencies in source code (see extractLuaDependencies).

Attention! This method clears all previous errors! See also docs of getErrors()

Return

a pointer to the created object or nullptr if something went wrong during creation. In that case, use getErrors() to obtain errors. The script can be destroyed by calling the destroy method

Parameters

  • source: the Lua source code

  • config: configuration options, e.g. for module dependencies

  • scriptName: name to assign to the script once it’s created

LuaModule *createLuaModule(std::string_view source, const LuaConfig &config = {}, std::string_view moduleName = "")

Creates a new rlogic::LuaModule from Lua source code. LuaModules can be used to share code and data constants across scripts or other modules. See also createLuaScript and rlogic::LuaConfig for details. You can optionally provide Lua module dependencies via the config, they will be accessible under their configured alias name for use by the module. The provided module dependencies must exactly match the declared dependencies in source code (see extractLuaDependencies).

Attention! This method clears all previous errors! See also docs of getErrors()

Return

a pointer to the created object or nullptr if something went wrong during creation. In that case, use getErrors() to obtain errors. The script can be destroyed by calling the destroy method

Parameters

  • source: module source code

  • config: configuration options, e.g. for module dependencies

  • moduleName: name to assign to the module once it’s created

bool extractLuaDependencies(std::string_view source, const std::function<void(const std::string&)> &callbackFunc)

Extracts dependencies from a Lua script or module source code so that the corresponding modules can be provided when creating rlogic::LuaScript or rlogic::LuaModule.

Any rlogic::LuaScript or rlogic::LuaModule which has a module dependency, i.e. it requires another rlogic::LuaModule for it to work, must explicitly declare these dependencies directly in their source code by calling function ‘modules’ in global space and pass list of module names it depends on, for example: 

modules("foo", "bar")
function interface()
  OUT.x = foo.myType()
end
function run()
  OUT.x = bar.doSth()
end
The ‘modules’ function does not affect any other part of the source code in any way, it is used only for the purpose of explicit declaration and extraction of its dependencies.

Please note that script runtime errors are ignored during extraction. In case a runtime error prevents the ‘modules’ function to be called, this method will still succeed but will not extract any modules, i.e. will not call callbackFunc. It is therefore highly recommended to put the modules declaration always at the beginning of every script before any other code so it will get executed even if there is runtime error later in the code.

Attention! This method clears all previous errors! See also docs of getErrors()

Return

true if extraction succeeded (also if no dependencies found) or false if something went wrong. In that case, use getErrors() to obtain errors.

Parameters

  • source: source code of module or script to parse for dependencies

  • callbackFunc: function callback will be called for each dependency found

RamsesNodeBinding *createRamsesNodeBinding(ramses::Node &ramsesNode, ERotationType rotationType = ERotationType::Euler_XYZ, std::string_view name = "")

Creates a new rlogic::RamsesNodeBinding which can be used to set the properties of a Ramses Node object. The initial values of the binding’s properties are loaded from the ramsesNode. Rotation values are taken over from the ramsesNode only if the conventions are compatible (see rlogic::ERotationType).

Attention! This method clears all previous errors! See also docs of getErrors()

Return

a pointer to the created object or nullptr if something went wrong during creation. In that case, use getErrors() to obtain errors. The binding can be destroyed by calling the destroy method

Parameters

  • ramsesNode: the ramses::Node object to control with the binding.

  • rotationType: the type of rotation to use (will affect the ‘rotation’ property semantics and type).

  • name: a name for the new rlogic::RamsesNodeBinding.

RamsesAppearanceBinding *createRamsesAppearanceBinding(ramses::Appearance &ramsesAppearance, std::string_view name = "")

Creates a new rlogic::RamsesAppearanceBinding which can be used to set the properties of a Ramses Appearance object.

Attention! This method clears all previous errors! See also docs of getErrors()

Return

a pointer to the created object or nullptr if something went wrong during creation. In that case, use getErrors() to obtain errors. The binding can be destroyed by calling the destroy method

Parameters

RamsesCameraBinding *createRamsesCameraBinding(ramses::Camera &ramsesCamera, std::string_view name = "")

Creates a new rlogic::RamsesCameraBinding which can be used to set the properties of a Ramses Camera object.

Attention! This method clears all previous errors! See also docs of getErrors()

Return

a pointer to the created object or nullptr if something went wrong during creation. In that case, use getErrors() to obtain errors. The binding can be destroyed by calling the destroy method

Parameters

template<typename T>
DataArray *createDataArray(const std::vector<T> &data, std::string_view name = "")

Creates a new rlogic::DataArray to store data which can be used with animations. Provided data must not be empty otherwise creation will fail. See rlogic::CanPropertyTypeBeStoredInDataArray and rlogic::PropertyTypeToEnum to determine supported types that can be used to create a rlogic::DataArray.

Attention! This method clears all previous errors! See also docs of getErrors()

Return

a pointer to the created object or nullptr if something went wrong during creation. In that case, use getErrors() to obtain errors.

Parameters

AnimationNode *createAnimationNode(const AnimationChannels &channels, std::string_view name = "")

Creates a new rlogic::AnimationNode for animating properties. Refer to rlogic::AnimationNode for more information about its use. There must be at least one channel provided, please see rlogic::AnimationChannel requirements for all the data.

Attention! This method clears all previous errors! See also docs of getErrors()

Return

a pointer to the created object or nullptr if something went wrong during creation. In that case, use getErrors() to obtain errors.

Parameters

  • channels: list of animation channels to be animated with this animation node.

  • name: a name for the the new rlogic::AnimationNode.

TimerNode *createTimerNode(std::string_view name = "")

Creates a new rlogic::TimerNode for generate and/or propagate timing information. Refer to rlogic::TimerNode for more information about its use.

Attention! This method clears all previous errors! See also docs of getErrors()

Return

a pointer to the created object or nullptr if something went wrong during creation. In that case, use getErrors() to obtain errors.

Parameters

bool update()

Updates all rlogic::LogicNode’s which were created by this LogicEngine instance. The order in which rlogic::LogicNode’s are executed is determined by the links created between them (see link and unlink). rlogic::LogicNode’s which don’t have any links between then are executed in arbitrary order, but the order is always the same between two invocations of update without any calls to link or unlink between them. As an optimization rlogic::LogicNode’s are only updated, if at least one input of a rlogic::LogicNode has changed since the last call to update. If the links between logic nodes create a loop, this method will fail with an error and will not execute any of the logic nodes.

Attention! This method clears all previous errors! See also docs of getErrors()

Return

true if the update was successful, false otherwise In case of an error, use getErrors() to obtain errors.

void enableUpdateReport(bool enable)

Enables collecting of statistics during call to update which can be obtained using getLastUpdateReport. Once enabled every subsequent call to update will be instructed to collect various statistical data which can be useful for profiling and optimizing the network of logic nodes. Note that when enabled there is a slight performance overhead to collect the data, it is recommended to use this only during a development phase.

Parameters

  • enable: true or false to enable or disable update reports.

LogicEngineReport getLastUpdateReport() const

Returns collection of statistics from last call to update if reporting is enabled (enableUpdateReport). The report contains lists of logic nodes that were executed and not executed and other useful data collected during last update. See rlogic::LogicEngineReport for details. The report data is generated only if previously enabled using enableUpdateReport and is empty otherwise. The data is only relevant for the last update and is overwritten during next update. Note that if update fails the report contents are undefined.

Return

collected statistics from last update.

void setStatisticsLoggingRate(size_t loggingRate)

Set the logging rate, i.e. how often statistics will be logged. Logging rate of N means every Nth call to update statistics will be logged. Whether the the logs are actually logged is also influenced by the statistics log level that can be set with setStatisticsLogLevel. The logging rate also determines how many collected sets will be used to calculate min/max and average. These statistics include:

  • Time since last log in seconds

  • Update execution time in microseconds (Avg, Min, Max)

  • Time between update calls in microseconds (Avg, Min, Max)

  • Count of nodes executed in percentage of total count (Avg, Min, Max)

  • Links activated (Avg, Min, Max) When loggingRate is set to 0 the logging of statistics is disabled. Note that there is a slight performance overhead for collecting the statistics data, however on most platforms this should be marginal. To get more detailed information about update execution timings see getLastUpdateReport.

Parameters

  • loggingRate: rate of N means statistics will be logged every Nth call to update. By default loggingRate is 60.

void setStatisticsLogLevel(ELogMessageType logLevel)

Update statistics default logLevel is ELogMessageType::Debug. For the statistics to be logged the logLevel has to be <= then the result returned from rlogic::Logger::GetLogVerbosityLimit. For example if rlogic::Logger::GetLogVerbosityLimit returns ELogMessageType::Info you have to set statistics logLevel to ELogMessageType::Info or a smaller level (e.g. ELogMessageType::Warn) to display statistics. Note that setting the statistics log level only influences the periodic statistic logs. All other logs are not influenced by this method. To control the rate after how many updates a log is produced refer to setStatisticsLoggingRate.

Parameters

  • logLevel: the logLevel of statistics messages

bool link(const Property &sourceProperty, const Property &targetProperty)

Links a property of a rlogic::LogicNode to another rlogic::Property of another rlogic::LogicNode. After linking, calls to update will propagate the value of sourceProperty to the targetProperty. Creating links influences the order in which scripts are executed - if node A provides data to node B, then node A will be executed before node B. A single output property (sourceProperty) can be linked to any number of input properties (targetProperty), but any input property can have at most one link to an output property (links are directional and support a 1-to-N relationships).

The link() method will fail when:

Creating links which form a cycle in the node dependency graph will not fail right away but during the next call to update(). Links are directional, it is OK to have A->B, A->C and B->C, but is not OK to have A->B->C->A. Cycles are allowed only under special conditions when using a weak link (see linkWeak).

After calling link, the value of the targetProperty will not change until the next call to update. Creating and destroying links generally has no effect until update is called.

Attention! This method clears all previous errors! See also docs of getErrors()

Return

true if linking was successful, false otherwise. To get more detailed error information use getErrors()

Parameters

  • sourceProperty: the output property which will provide data for targetProperty

  • targetProperty: the target property which will receive its value from sourceProperty

bool linkWeak(const Property &sourceProperty, const Property &targetProperty)

A weak link is for the most part equivalent to link - it has the same requirements and data propagation behavior, but with one crucial difference. Weak link is not considered for node dependency graph (logic node topology) which determines the order of nodes execution during update. This allows to form a cycle using weak link within the dependency graph, which can be helpful when some feedback data is required (e.g. a controller node managing a worker node using normal links, worker node reports progress back to controller via weak link). However there are important things to consider to avoid unexpected problems:

  1. The value propagated from a weak link is old, it is the value that was calculated in previous update. Example: consider A->B->C using normal links and a back loop C->A using weak link. When update is called, the nodes are executed exactly in the order defined by normal link dependency: A, B, C (remember that weak links are not considered for execution order). This means however that when A is executed its input value from weak linked C is only from previous update because C is executed last.

  2. During first update (after creation or load from file) there is no value applied from weak link, instead the initial input property value will be used at the weak link target. This is a logical consequence of the limitation described above - during first update there was no previous update and therefore no value from weak link.

  3. Avoid infinite update loops - update has an optimization to execute only nodes with modified inputs, this can greatly simplify update of complex node hierarchies. When using weak links there is a risk that there will be a never ending need of executions of nodes involved in a node graph cycle. Consider the last node providing new value via weak link to first node, which then generates again new value propagated to the last node, this will result in the need to execute those nodes in all upcoming update calls. Try to avoid such case and if it is not possible at least try to limit this update loop to a minimum of nodes.

Attention! This method clears all previous errors! See also docs of getErrors()

Return

true if linking was successful, false otherwise. To get more detailed error information use getErrors()

Parameters

  • sourceProperty: the output property which will provide data for targetProperty

  • targetProperty: the target property which will receive its value from sourceProperty

bool unlink(const Property &sourceProperty, const Property &targetProperty)

Unlinks two properties which were linked with link. After a link is destroyed, calls to update will no longer propagate the output value from the sourceProperty to the input value of the targetProperty. The value of the targetProperty will remain as it was after the last call to update - it will not be restored to a default value or to any value which was set manually with calls to rlogic::Property::set().

Attention! This method clears all previous errors! See also docs of getErrors()

Return

true if unlinking was successful, false otherwise. To get more detailed error information use getErrors().

Parameters

  • sourceProperty: the output property which is currently linked to targetProperty

  • targetProperty: the property which will no longer receive the value from sourceProperty

bool isLinked(const LogicNode &logicNode) const

Checks if an input or output of a given LogicNode is linked to another LogicNode

Return

true if the given LogicNode is linked to any other LogicNode, false otherwise.

Parameters

  • logicNode: the node to check for linkage.

const std::vector<ErrorData> &getErrors() const

Returns the list of all errors which occurred during the last API call to a LogicEngine method or any other method of its subclasses (scripts, bindings etc). Note that errors get wiped by all mutable methods of the LogicEngine.

This method can be used in two different ways:

  • To debug the correct usage of the Logic Engine API (e.g. by wrapping all API calls with a check for their return value and using this method to find out the cause of the error)

  • To check for runtime errors of scripts which come from a dynamic source, e.g. by calling the method after an unsuccessful call to update() with a faulty script

Return

a list of errors

bool destroy(LogicObject &object)

Destroys an instance of an object created with LogicEngine. All objects created using LogicEngine derive from a base class rlogic::LogicObject and can be destroyed using this method.

In case of a rlogic::LogicNode and its derived classes, if any links are connected to this rlogic::LogicNode, they will be destroyed too. Note that after this call, the execution order of rlogic::LogicNode may change! See the docs of link and unlink for more information.

In case of a rlogic::DataArray, destroy will fail if it is used in any rlogic::AnimationNode’s rlogic::AnimationChannel.

In case of a rlogic::LuaModule, destroy will fail if it is used in any rlogic::LuaScript.

Attention! This method clears all previous errors! See also docs of getErrors()

Return

true if object destroyed, false otherwise. Call getErrors() for error details upon failure.

Parameters

  • object: the object instance to destroy

bool saveToFile(std::string_view filename)

Writes the whole LogicEngine and all of its objects to a binary file with the given filename. The RAMSES scene potentially referenced by rlogic::RamsesBinding objects is not saved - that is left to the application. LogicEngine saves the references to those object, and restores them after loading. Thus, deleting Ramses objects which are being referenced from within the LogicEngine will result in errors if the Logic Engine is loaded from the file again. Note that it is not sufficient to have objects with the same name, they have to be the exact same objects as during saving! For more in-depth information regarding saving and loading, refer to the online documentation at https://ramses-logic.readthedocs.io/en/latest/api.html#saving-loading-from-file

Note: The method reports error and aborts if the rlogic::RamsesBinding objects reference more than one Ramses scene (this is acceptable during runtime, but not for saving to file).

Attention! This method clears all previous errors! See also docs of getErrors()

Return

true if saving was successful, false otherwise. To get more detailed error information use getErrors()

Parameters

  • filename: path to file to save the data (relative or absolute). The file will be created or overwritten if it exists!

bool loadFromFile(std::string_view filename, ramses::Scene *ramsesScene = nullptr, bool enableMemoryVerification = true)

Loads the whole LogicEngine data from the given file. See also saveToFile(). After loading, the previous state of the LogicEngine will be overwritten with the contents loaded from the file, i.e. all previously created objects (scripts, bindings, etc.) will be deleted and pointers to them will be invalid. The (optionally) provided ramsesScene will be used to resolve potential rlogic::RamsesBinding objects which point to Ramses objects. You can provide a nullptr if you know for sure that the LogicEngine loaded from the file has no rlogic::RamsesBinding objects which point to a Ramses scene object. Otherwise, the call to loadFromFile will fail with an error. In case of errors, the LogicEngine may be left in an inconsistent state. For more in-depth information regarding saving and loading, refer to the online documentation at https://ramses-logic.readthedocs.io/en/latest/api.html#saving-loading-from-file

Attention! This method clears all previous errors! See also docs of getErrors()

Return

true if deserialization was successful, false otherwise. To get more detailed error information use getErrors()

Parameters

  • filename: path to file from which to load content (relative or absolute)

  • ramsesScene: pointer to the Ramses Scene which holds the objects referenced in the Ramses Logic file

  • enableMemoryVerification: flag to enable memory verifier (a flatbuffers feature which checks bounds and ranges). Disable this only if the file comes from a trusted source and performance is paramount.

bool loadFromBuffer(const void *rawBuffer, size_t bufferSize, ramses::Scene *ramsesScene = nullptr, bool enableMemoryVerification = true)

Loads the whole LogicEngine data from the given memory buffer. This method is equivalent to loadFromFile but allows to have the file-opening logic done by the user and only pass the data as a buffer. The logic engine only reads the data, does not take ownership of it and does not modify it. The memory can be freed or modified after the call returns, the LogicEngine keeps no references to it.

Return

true if deserialization was successful, false otherwise. To get more detailed error information use getErrors()

Parameters

  • rawBuffer: pointer to the raw data in memory

  • bufferSize: size of the data (bytes)

  • ramsesScene: pointer to the Ramses Scene which holds the objects referenced in the Ramses Logic file

  • enableMemoryVerification: flag to enable memory verifier (a flatbuffers feature which checks bounds and ranges). Disable this only if the file comes from a trusted source and performance is paramount.

LogicEngine(const LogicEngine &other) = delete

Copy Constructor of LogicEngine is deleted because logic engines hold named resources and are not supposed to be copied

Parameters

  • other: logic engine to copy from

LogicEngine(LogicEngine &&other) noexcept

Move Constructor of LogicEngine

Parameters

  • other: logic engine to move from

LogicEngine &operator=(const LogicEngine &other) = delete

Assignment operator of LogicEngine is deleted because logic engines hold named resources and are not supposed to be copied

Parameters

  • other: logic engine to assign from

LogicEngine &operator=(LogicEngine &&other) noexcept

Move assignment operator of LogicEngine

Parameters

  • other: logic engine to move from

Public Members

std::unique_ptr<internal::LogicEngineImpl> m_impl

Implementation detail of LogicEngine