Subject: Re: [boost] Finite state machine (arcolight)
From: arcolight (arcolight_at_[hidden])
Date: 2014-09-07 14:15:50
> This is something that would be very interesting to me, but I wonder if the
> memory allocation can be factored out in the current boost library by using
> allocators or something. I'm not terrible familiar with the libraries
> internals, but I'd be willing to take a look at my next free cycle. To be
> clear, is there another state machine lib in boost besides
> boost.statechart? That's the only one I'm familiar with from a user
Yes, custom allocator doesn't solve the problem, especially because current FSM-s includes many other libraries. In any case, I can't afford to use even custom allocator in my project, if I would use it, I must have sufficient size heap, it's risky in embedded system. I already have third-party code, that uses heap, and I couldn't do anything with it. :)
Another FSM implementation - Boost.MSM(Meta state machine), I tried to use it in different ways, but i've failed. MSM based on MPL boost library.
If you interested in code, I have small repo on github: https://github.com/arcolight/ecl, there are several small classes, one of them, in fsm.hpp - state machine. There are simple test with cd-player like in boost lib.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk