Confusion about error.execution #644

The error.execution is causing me some confusion.

1. Is it true that when defining the error_execution event in my machine, I must manually give it an id manually usingEvent(..., id="error.execution")?

   Before sending an event to the state machine, I check whether the event can be triggered using sm.allowed_events. Then I needed to catch all unhandled exceptions that could be raised by the listeners (StateChart itself, Model, etc.). I learned from the Error Handling page that this can be done by adding handlers for the error.execution event. But it seems sm.allowed_events and sm.enabled_events() conflicts with the error_execution event.

   I will demonstrate this using the ErrorLogger example from the Error Handling section:

   from statemachine import State, StateChart
   class ErrorLogger(StateChart):
    running = State(initial=True)
    failed = State(final=True)
    process = running.to(running, on="do_process")
    error_execution = running.to(failed, on="log_error")
    def do_process(self):
    raise ValueError("bad data")
    def log_error(self, error):
    self.last_error = error
   

   When I try to get enabled events after the machine reaches a stable configuration:

   sm = ErrorLogger()
   print(f" Enabled events: {sm.enabled_events()}")
   

   I get the AttributeError exception:

   (.venv) PS <...>> sm-test.exe
   Traceback (most recent call last):
   File "<frozen runpy>", line 203, in _run_module_as_main
   File "<frozen runpy>", line 88, in _run_code
   File "<...>\.venv\Scripts\sm-test.exe\__main__.py", line 5, in <module>
    sys.exit(main())
    ~~~~^^
   File "<...>\src\sm_test\__main__.py", line 57, in main
    print(f" Enabled events: {sm.enabled_events()}")
    ~~~~~~~~~~~~~~~~~^^
   File "<...>\.venv\Lib\site-packages\statemachine\statemachine.py", line 450, in enabled_events
    result = self._engine.enabled_events(*args, **kwargs)
   File "<...>\.venv\Lib\site-packages\statemachine\engines\sync.py", line 191, in enabled_events
    "event": getattr(sm, event),
    ~~~~~~~^^^^^^^^^^^
   AttributeError: 'ErrorLogger' object has no attribute 'error.execution'. Did you mean: 'error_execution'?
   

   I found that the problem disappears if I manually assign the name error.execution to the error_execution event:

   class ErrorLogger(StateChart):
    [...]
    error_execution = Event(running.to(failed, on="log_error"), id="error.execution")
   

2. Is it true that error_execution event, by design, can be sent to the machine from "outside"?

   When I applied the workaround with Event(..., id="error.execution"), the sm.enabled_events() gives:

   Enabled events: [BoundEvent('process', delay=0, internal=False), BoundEvent('error.execution', delay=0, internal=False)]
   

   Note that error.execution is enabled event, and it can be sent to the machine from main().

   This snippet:

   sm = ErrorLogger()
   print(f" Enabled events: {sm.enabled_events()}")
   if "error.execution" in sm.enabled_events():
    sm.send("error.execution", error=RuntimeError("Should this be possible?"))
   print(sm.last_error)
   

   gives:

    Enabled events: [BoundEvent('process', delay=0, internal=False), BoundEvent('error.execution', delay=0, internal=False)]
   Should this be possible?
   

   The error.execution event is described as an internal mechanism of the machine, and it’s strange that this event can be sent from "outside".

python-statemachine versions 3.0.0, 3.1.0, 3.2.0, and 3.2.1 work identically with the specified code snippets.