On Mi, Sep 18, 2013 at 02:58:46 +0200, Ruben Smits wrote:
> Hi (Markus),
> I just installed guard functions on some of my transitions of a RFSM state
> machine. I expected that the guard function would only be executed when the
> transition was triggered by the installed event. But apparently the guards are
> executed at each run of the fsm even when the event that triggers the transition
> was not raised. Is this the desired behavior or a bug?

It's desired, the guard is checked before the 'events' table
specification. As the guard is merely allowed to be a side-effect free
predicate function, the order does not matter, hence you should not
(have to) rely on it. If you need to do sth. during a transition, you
should use the 'effect' instead.

Markus

PS: Note that the 'events' table is merely a syntactic convenience,
the guard by itself would suffice to express the same behaviour. I
have a trigger module in my queue that allows to specify more complex
conditions, including AND, negation, XOR, over multiple
steps. Internally these are translated to an extended guard function.