Re: [Boost-users] MSM accepts events before calling start

Subject: Re: [Boost-users] MSM accepts events before calling start
From: Christophe Henry (christophe.j.henry_at_[hidden])
Date: 2012-08-30 16:45:06

Hi Andreas,

> Dear all
>
> Recently I discovered MSM - a very nice piece of code!

Thanks :)

> It let me define the state machine decoupled from the other business logic 
> code
> what brought me easier unit testing and easier life :-)
>
> Really nice so far. This week I faced the following situation: my MSM
> accepts and processes events BEFORE its  start() method was called. Is
> this behavior by design? I know other state machine implementations
> which do not accept any events until they are started. If the MSM's
> behavior related to this start issue is on purpose, can anybody tell
> me the background? Just curious if others have same thoughts or if my
> mind acrobatic is totally odd.

This is by design. Actually it just means "call first state entry action 
(initial state)" but it was more readable than "call_inital_state_entry".
The rationale is, MSM actually doesn't need a start(), but first MSM users 
complained that MSM was calling on_entry on initial state upon creation and 
they wanted to choose the time themselves. I although thought I could do 
other things in start() one day but I never did.

The same way, you also have a stop(), which calls on_exit on the last active 
state(s).

You can omit both, simply on_entry of initial state(s) and on_exit of the 
last active state(s) (and of the state machine itself) will not be called.

This means that most of the time, if you forget start(), nothing bad 
happens.

> Thanks for any comments
>
>
> Andi

HTH,
Christophe