0

I am trying to build a DSL for my iOS project.
For this I plan to build a Semantic Model in a form of a State Machine. (The terminology is from Fowler's book on DSLs).
The main idea: State Machine is coded as the set of states and transitions between them, and then some unit tests can be written to check if some_action on some_state1 leads the system to a some_state2.

The problem is my app has a lot of background threads, so more than one state can be active in State Machine at a given moment.

I've read that such State Machines are Nondeterministic finite automaton, skimmed through wiki page, but that looks too theoretical for me.

Here is an example state machine:

s8 can be activated only when it received t7 and t8, that means it should "wait".

enter image description here


The questions:
1. Is there something like "wait" in state machines?
2. Maybe, this is not NFA, but two state machines? Should I care how to name this kind of semantic model at all?
3. Is it OK to s8 be implemented with some background thread, which accepts notifications from s4 and s7, and is activated only when both of them sent notifications (this means that the unit test fail when timeout is reached, and then that timeout should be mentioned somewhere in a model) ?

olha
  • 2,132
  • 1
  • 18
  • 39

1 Answers1

1

Take a look at orthogonal regions in a hierarchical state machine, this should do what you are looking for.

Create a new state with two child regions: the first region containing s1, s2, s3, s4 and a new final state (the target of t8); the second region containing s5, s6, s7 and a new final state (the target of t7). Then add a new transition from the new state to s8.

The new transition should only be traversed when both regions are in a final state.

Mesmo
  • 1,890
  • 2
  • 15
  • 13