[boost] Re: ANN: review of FSM library starts

From: Mick Hollins (mick_at_[hidden])
Date: 2005-03-02 17:11:41

Andreas Huber wrote:
> 
>>In general, I would drop all mention of UML except for some appendix
>>somewhere (and the reference manual too).
> 
> 
> Why?

UML is a widely accepted mechanism for describing state machines, so why shouldn't the tutorial make use of them?

I (very) recently used boost::fsm for the first time and I think arguably its best feature is the fact that the mapping 
from a UML state-chart diagram to boost::fsm code is trivial. That allows me to design my state machine graphically 
(which my brain prefers) and to then perform the transformation to code as a trivial step. So, in my opinion, 
boost::fsm's relationship to UML is one of its primary features which deserves to be emphasised, not whacked in an appendix.

I might post a longer review if I get the chance, but my basic summary is that boost::fsm is far and away the best C/C++ 
library I've seen for implementing finite state machines simply and reliably. It has literally been a godsend for some 
of the stuff I am trying to achieve.

So, a big thanks to Andreas, and whether boost::fsm makes it into boost or not, I'll still be using it :-)

cheers,
mick