Skip to content

5.4.2.3.3. Fork

Mikołaj Milewski edited this page Nov 12, 2026 · 1 revision

Fork

Fork is a type of Pseudostate used in State Machines to model concurrent transitions. It allows the State Machine to split its flow into multiple parallel paths, each leading to a different target state.

Each Transition outgoing from a Fork must be targeted in a State or Pseudostate defined in different orthogonal Region of Orthogonal State.

All Transitions coming out from a Fork are Default Transitions - they are not explicitly triggered by an Event.

As Fork is a Pseudostate and not a State, the State Machine cannot stop its flow on it. Therefore, no Guards are allowed in Transitions outgoing from a Fork.

Definition

In UML, a Fork pseudostate is represented by a horizontal or vertical bar with multiple outgoing transitions:

stateDiagram-v2
state Fork <<fork>>
[*] --> InitialState
InitialState --> Fork : TriggerEvent
state ParallelStates {
    State1
    --
    State2
    --
    State3
}
Fork --> State1
Fork --> State2
Fork --> State3
Loading

Equivalent Stateflows notation of a Fork pseudostate:

Lambda style

Definition fragment (inside the builder callback):

    /* fragment of State Machine definition */
    .AddInitialState("InitialState", b => b
        .AddTransition<TriggerEvent>("Fork")
    )
    .AddFork("Fork", b => b
        .AddTransition("State1")
        .AddTransition("State2")
        .AddTransition("State3")
    )
    .AddOrthogonalState("ParallelStates", parallel => parallel
        .AddRegion(region => region.AddInitialState("State1"))
        .AddRegion(region => region.AddInitialState("State2"))
        .AddRegion(region => region.AddInitialState("State3")))

If there is just one Fork in the State Machine, a shortened notation with the default name can be used:

Unnamed pseudostate shorthand

Definition fragment (inside the builder callback):

    /* fragment of State Machine definition */
    .AddInitialState("InitialState", b => b
        .AddTransition<TriggerEvent>(Fork.Name)
    )
    .AddFork(b => b // no name provided here, default name is used
        .AddTransition("State1")
        .AddTransition("State2")
        .AddTransition("State3")
    )
    .AddOrthogonalState("ParallelStates", parallel => parallel
        .AddRegion(region => region.AddInitialState("State1"))
        .AddRegion(region => region.AddInitialState("State2"))
        .AddRegion(region => region.AddInitialState("State3")))

Using lambda style means that the Fork pseudostate and its Transitions are defined using string names.

Typed style

Definition fragment (inside the builder callback):

    /* fragment of State Machine definition */
    .AddInitialState<InitialState>(b => b
        .AddTransition<TriggerEvent, Fork>()
    )
    .AddFork(b => b
        .AddTransition<State1>()
        .AddTransition<State2>()
        .AddTransition<State3>()
    )
    .AddOrthogonalState("ParallelStates", parallel => parallel
        .AddRegion(region => region.AddInitialState<State1>())
        .AddRegion(region => region.AddInitialState<State2>())
        .AddRegion(region => region.AddInitialState<State3>()))

Using typed style here means that the Fork pseudostate is represented by a class, in the example predefined Fork class is used.

Fork class can be used as a target for typed style Transitions if the State Machine contains just one Fork. If there are more Forks, there is an important concern of State identity to be considered.

Clone this wiki locally