From: ara.t.howard@...
Date: 2006-09-10T05:27:20+09:00
Subject: Re: state machine in ruby

On Sun, 10 Sep 2006, Francis Cianfrocca wrote:

> You'd have to add a parse stack and a lookahead  buffer.  Might not be that
> easy, but I'll take a look if I get a chance.

k.  go easy on me.  it's version 0.0.0 ;-)

>> > And given that text, mechanically generate a state machine that would be
>> > restartable so it could be event-driven. And the scanner would be pretty
>> > funky too, it would have to be able to generate a token that means "not
>> > enough input to scan a complete token, go to sleep now."
>> 
>> naw, that's easy.  just do
>
>
> You solved that?? Nice.

well - the event part:

     harp:~ > cat a.rb
     require "time"
     require "fsm"

     STDOUT.sync = true
     Thread.abort_on_exception = true

     fsys = FSM.system{
       fsm{
         states %w( a b c )
         transition "a2b", "a" => "b"
         transition "b2c", "b" => "c"
       }
       #start # leave starting state and enter state "a"
     }

     Thread.new{
       sleep 1
       fsys.start # enters state 'a'
       sleep 1
       fsys.fsm.transition "a", "a2b"
       sleep 1
       fsys.fsm.transition "b", "b2c"
     }

     Thread.new{
       puts "waiting for state a @ #{ Time.now.iso8601(6) }..."
       fsys.observer.wait_for "a"
       puts "state a waited for."

       puts "waiting for state b @ #{ Time.now.iso8601(6) }..."
       fsys.observer.wait_for "b"
       puts "state b waited for."

       puts "waiting for state c @ #{ Time.now.iso8601(6) }..."
       fsys.observer.wait_for "c"
       puts "state c waited for."
     }.join


     harp:~ > ruby a.rb
     waiting for state a @ 2006-09-09T14:03:42.527586-06:00...
     state a waited for.
     waiting for state b @ 2006-09-09T14:03:43.537855-06:00...
     state b waited for.
     waiting for state c @ 2006-09-09T14:03:44.547454-06:00...
     state c waited for.


there a lots of other event hooks too.  wait_for_once.  etc.  regarding the
special token, that's easy - there are also hooks for input into the system:

     harp:~ > cat a.rb
     require "time"
     require "fsm"

     STDOUT.sync = true
     Thread.abort_on_exception = true
     MSGS = Queue.new and Thread.new{ loop{ puts MSGS.pop } }
     def t() Time.now.iso8601(6) end
     def p(msg) MSGS.push msg end


     fsys = FSM.system{
       fsm{
         states %w( a b c )
         transition "a2b", "a" => "b"
         transition "b2c", "b" => "c"
         transition "c2a", "c" => "a"
       }

       observer{
         on_entry do |info|
           p "entry <#{ info.inspect }> @ #{ t }"
         end
         on_input do |*argv|
           p "input <#{ argv.inspect }> @ #{ t }"
         end
       }
       start
     }


     Thread.new{
       1.times do
         sleep 1
         fsys.input 'A'
         fsys.transition "a", "a2b"
         sleep 1

         sleep 1
         fsys.input 'B'
         fsys.transition "b", "b2c"
         sleep 1

         sleep 1
         fsys.input 'C'
         fsys.transition "c", "c2a"
         sleep 1
       end
     }.join



     jib:~ > ruby a.rb
     entry <[FSM::Event::Entry, "a", nil]> @ 2006-09-09T14:23:43.335322-06:00
     input <[[FSM::Event::Input, "a", "A"]]> @ 2006-09-09T14:23:44.338837-06:00
     entry <[FSM::Event::Entry, "b", nil]> @ 2006-09-09T14:23:44.340079-06:00
     input <[[FSM::Event::Input, "b", "B"]]> @ 2006-09-09T14:23:46.358668-06:00
     entry <[FSM::Event::Entry, "c", nil]> @ 2006-09-09T14:23:46.359932-06:00
     input <[[FSM::Event::Input, "c", "C"]]> @ 2006-09-09T14:23:48.378684-06:00
     entry <[FSM::Event::Entry, "a", nil]> @ 2006-09-09T14:23:48.380190-06:00


so, for the special token, you simply remain in the current state - the
observer is already sleeping ;-)

is this what you are thinking of?


>> what's AMQP ??
>
> Advanced Message Queueing Protocol. Very interesting if you're into
> message-queueing, it's the evolution of the "amq" that J.P.Morgan has been
> working on for a couple of years and that you've heard periodic rumors about
> during that time. It's different because it specifies not only the wire
> protocol (which itself is state-of-the-art and worth a look) but also the
> broker architecture. Needless to say, all the current work has been done in
> Java, so they need an agile implementation. :-)

learn something every day!

cheers.

-a
-- 
what science finds to be nonexistent, we must accept as nonexistent; but what
science merely does not find is a completely different matter... it is quite
clear that there are many, many mysterious things.
- h.h. the 14th dalai lama