Subj : Re: implementation of finite state machine
To   : comp.programming
From : moi
Date : Thu Oct 06 2005 08:58 pm

Thad Smith wrote:
> pete wrote:
> 
>>Roman Mashak wrote:
>>
>>>Hello, Dave!
>>>You wrote  on Thu, 06 Oct 2005 10:11:59 +0100:
>>>
>>> ??>> Could you please point me to some resource, describing sample
>>> ??>> implementation of FSM in C language, just to better catch
>>> ??>> idea. TIA. With best regards, Roman Mashak.  E-mail: mrv@tusur.ru
>>>

>>>Due to my understandanding of FSM,
>>>I used 'if-elseif-else' chain. Might me
>>>there are other ways and more efficient methods?
>>
>>Use an array of function pointers.
>>The index into the array, is the state.
>>The state is updated either as the return value of the function call
>>or the address of the state object can be passed as an argument.
> 
> 
> Another common technique is a switch statement, with a case for each
> state.  The case logic evaluates input criteria to determine when to
> change state.
> 
> Another technique is table driven, based on small number of input
> variables.  Index into the array based on input values to choose the
> next state.
> 
> In both techniques you can use either a particular transition or
> particular state to choose an action routine (or block of code) to
> perform based on either the transition or the current state.
> 
> Thad

An axample of the "next state := big_table [ current state, token ] "
-method can be found in [f]lex, which generates the big tables for you.

HTH,
AvK

.