Hierarchical States Machines

Hierarchical States Machines

[Fabien]

Hi everyone !!!

Pretty new to ADA dev., i am currently trying to implement a
hierarchical state machine for my application. So far, i must confess
that i do not really get on well with it. I am thus looking for some
people who might have tried to do it and who could give me some useful
tips .

Fabien

pS : I don't look for a ready-to-use solution, just some help !!!!!

Re: Hierarchical States Machines

[Marius Amado Alves]

> Pretty new to ADA dev., 

Welcome! You can now start writing "Ada" not "ADA" as your first rite of 
passage :-)

> i am currently trying to implement a
> hierarchical state machine for my application....

I write state machines a lot in Ada. For example package XML_Automaton, which 
I've just put for your viewing pleasure at the following secret URLs:

http://www.liacc.up.pt/~maa/mneson/xml_automaton.ads
http://www.liacc.up.pt/~maa/mneson/xml_automaton.adb
http://www.liacc.up.pt/~maa/mneson/xml2mntext.adb

The last unit is an real life example application of the package.

(Soon I'll update the Mneson site with *all* the cool latest developments. As 
you might guess from above, XML import/export is one of them.)

Re: Hierarchical States Machines

[Robert I. Eachus]

Fabien wrote:

> Pretty new to ADA dev., i am currently trying to implement a
> hierarchical state machine for my application. So far, i must confess
> that i do not really get on well with it. I am thus looking for some
> people who might have tried to do it and who could give me some useful
> tips .
> 
> Fabien
> 
> pS : I don't look for a ready-to-use solution, just some help !!!!!

Unlike any other type of software, goto is your friend.

The reason that goto is normally deprecated is that it results in 
spaghetti code.  However, if you are implementing a state machine, you 
need to reflect the structure of the actual state diagram, which may 
look like spaghetti.

Also if you are implementing a table driven state machine, the best way 
to do it is to build a driver that is its own state machine.  (Sort of 
like using a universal Turing machine to read a description of a Turing 
machine and emulate it.)  Again, the use of gotos will allow you to 
write a driver so that its structure corresponds to the algorithm you 
are implementing.

Incidentally, as far as I know, or for that matter most Ada experts 
know, this is the only area in Ada where you should use gotos.  But goto 
has been kept in the language primarily because it is _necessary_ to 
create maintainable finite state machines, push-down automata, and the like.

-- 

                                           Robert I. Eachus

"The terrorist enemy holds no territory, defends no population, is 
unconstrained by rules of warfare, and respects no law of morality. Such 
an enemy cannot be deterred, contained, appeased or negotiated with. It 
can only be destroyed--and that, ladies and gentlemen, is the business 
at hand."  -- Dick Cheney

Re: Hierarchical States Machines

[Randy Brukardt]

"Robert I. Eachus" <rieachus@comcast.net> wrote in message
news:jrOdnQYzWKMxkQ3dRVn-hQ@comcast.com...
> Incidentally, as far as I know, or for that matter most Ada experts
> know, this is the only area in Ada where you should use gotos.  But goto
> has been kept in the language primarily because it is _necessary_ to
> create maintainable finite state machines, push-down automata, and the
like.

Ah, baloney. :-)

I use gotos whenever they prevent duplicating work that otherwise would have
to be done. It's silly to have two copies of code to do something just to
avoid a goto (or to make such code into a subprogram with no purpose other
than to avoid a goto).

A common case is to emulate the missing "continue" that Ada doesn't have. It
usually shows up in some deeply nested code:

    loop
        <bunch of conditions>
            <Do some processing>
            if <unusual condition> then
                  <Handle unusual case>
                  goto Continue;
           end if;
       <bunch of else conditions>
       end if;
       <normal processing>;
     <<continue>> null;
   end loop;

You can avoid the goto by making a hash out of the code, but why bother?
Gotos are efficient, not particularly unstructured in Ada (most of the bad
cases are illegal) - compiler optimizers are unlikely to be able to get rid
of batches of boolean flags and turn them into the branches that really were
intended in the first place.

                       Randy.

Re: Hierarchical States Machines

[Steve]

I don't have anything against using goto's for state machines... but...

I usually set up an enumeration for each state:

  type my_states is ( start_state, initializing, etc, end_state );

And then use a loop with an enclosed case statement:

  state := start_state;
  loop
    case state is
      when start_state =>
         ...
         state := initializing;
      when initializing =>
      when etc =>
      when end_state =>
    end case;
  end loop;

In my opinion it makes things easier to follow since the entry
to a new state always happens from the same place.

Steve
(The Duck)

"Fabien" <fab_lio@yahoo.fr> wrote in message
news:fed2d60.0404281048.41d6c549@posting.google.com...
> Hi everyone !!!
>
> Pretty new to ADA dev., i am currently trying to implement a
> hierarchical state machine for my application. So far, i must confess
> that i do not really get on well with it. I am thus looking for some
> people who might have tried to do it and who could give me some useful
> tips .
>
> Fabien
>
> pS : I don't look for a ready-to-use solution, just some help !!!!!

Re: Hierarchical States Machines

[Robert I. Eachus]

Steve wrote:

> I don't have anything against using goto's for state machines... but...
> 
> I usually set up an enumeration for each state...
> 
> In my opinion it makes things easier to follow since the entry
> to a new state always happens from the same place.

Exactly.  My point is that this:

loop
   case State is
     when...
   end case;
   -- post processing
end loop;

...is the real underlying structure, and you should use gotos when 
needed to preserve it.  Randy mentioned one such case continue. (His 
code looked exactly like a FSM to me, even if he wants to call it 
something else.)  Other typical patterns you run into are where you need 
  multiple exits from the loop, or you need to return to the beginning 
while skipping the post-processing.  For example in the Ada parser 
driver I wrote for state machines created by LALR1 on Multics, there 
were some "merged" state entries where several states action tables were 
almost identical.  So LALR1 generated table entries that said 
essentially: if the next token is X do Y, otherwise go to the state 
table for state Z.  You could distort the engine all out of shape to 
implement this, or put in a goto to just before the loop.

Someone who had been taught the "gotos are evil" religion reorganized 
the code with a couple of added state variables to demonstrate the goto 
wasn't necessary.  His code was only twenty lines longer than my code, 
and ran 40% slower. (Register pressure from those state variables push 
things that should have been in registers onto the stack.)  Although he 
talked about looking at the generated code to see why it was so much 
slower, I think he realized that his version was the spaghetti.

-- 

                                           Robert I. Eachus

"The terrorist enemy holds no territory, defends no population, is 
unconstrained by rules of warfare, and respects no law of morality. Such 
an enemy cannot be deterred, contained, appeased or negotiated with. It 
can only be destroyed--and that, ladies and gentlemen, is the business 
at hand."  -- Dick Cheney

Re: Hierarchical States Machines

[tmoran]

>and ran 40% slower. (Register pressure from those state variables push
  Do you mean the parser state transitions ran 40% slower, or the whole
compiler, including IO, lexer, symbol tables, code generation, etc
was slowed by 40%?

Re: Hierarchical States Machines

[Martin Krischik]

Randy Brukardt wrote:

>     loop
>         <bunch of conditions>
>             <Do some processing>
>             if <unusual condition> then
>                   <Handle unusual case>
>                   goto Continue;
>            end if;
>        <bunch of else conditions>
>        end if;
>        <normal processing>;
>      <<continue>> null;
>    end loop;

Is there any deepter reason why you did not use:

     loop
        <<continue>>
         <bunch of conditions>
             <Do some processing>
             if <unusual condition> then
                   <Handle unusual case>
                   goto Continue;
            end if;
        <bunch of else conditions>
        end if;
        <normal processing>;
    end loop;

i.E. A style guide or a RM I am not aware of?

With Regards

Martin

-- 
mailto://krischik@users.sourceforge.net
http://www.ada.krischik.com

Re: Hierarchical States Machines

[Wojtek Narczynski]

Hello,

> But goto has been kept in the language primarily because it is _necessary_ to
> create maintainable finite state machines, push-down automata, and the like.

In the absence of "last call optimisation".

Regards,
Wojtek

Re: Hierarchical States Machines

[Marius Amado Alves]

On Thursday 29 April 2004 04:58, Steve wrote:
> I don't have anything against using goto's for state machines... but...
>
> I usually set up an enumeration for each state:
>
>   type my_states is ( start_state, initializing, etc, end_state );...

That's how I usually do it too.

See

   http://www.liacc.up.pt/~maa/files/Ada_Automaton/

for an Ada tokenizer written this way. Programs Ada_Automaton_Test and 
Ada_Counter are example applications of the package. Ada_Counter counts the 
significant semicolons in a unit.

Re: Hierarchical States Machines

[Robert I. Eachus]

tmoran@acm.org wrote:
>>and ran 40% slower. (Register pressure from those state variables push
> 
>   Do you mean the parser state transitions ran 40% slower, or the whole
> compiler, including IO, lexer, symbol tables, code generation, etc
> was slowed by 40%?

Sorry, the first pass of the compiler, which included the scanner 
(lexer) and created the symbol table and AST.  The second pass did the 
semantic analysis and 'decorated' the parse tree, and did all the 
code-generator independent optimizations.

As I remember it, the running times were roughly 1-3-3.  So it would 
have been 1.4-3-3, for a 5 to 6% overall slowdown.

Even so it was pretty huge, and it shocked me.  By the point this 
occurred we had done a lot of optimization of the LALR1 engine and 
structure so that (mostly for error correcting reasons) successive 
reduce operations on the parse stack were deferred, then all called at 
once after two successful succeeding shifts, or a successful shift and 
(pending) reduce.

That sounds complicated, and it was.  But it made for wonderful syntax 
error correction without increasing the parse table size.  (Well we had 
about a half dozen added states to allow "panic mode" if none of the 
single or double token fixes worked.)

We were initially worried about the changes for the error correction 
support slowing down performance, but since it grouped the actions that 
resulted in subroutine calls, it actually sped things up significantly 
by eliminating register spills.  When we looked at the generated code to 
try and explain that 40% number, adding the two flags for the non-goto 
solution increased the number of active variables from 6 to 8 in the key 
loops--and we had six available registers once you eliminated the stack 
pointer and program counter.  Worse the access pattern basically rotated 
through the variables.  We thought about fixing that for both the Ada 
and Multics PL/1 compilers, but a little bit of monitoring indicated 
that this code was about the only thing around that hit this 'misfeature'.

The wonderful thing about being in a compiler group, and using Multics, 
was that we could slip in performance counters like this on the 
production machine, and see whether or not we should actually fix 
anything.  Technically our group supported the DPS-6 line not DPS-8M, 
but we were in Billerica, MA, Multics development was at CISL in 
Cambridge, MA, and GCOS-3/8 support was in Phoenix, AZ.  So I could 
submit a performance counter proposal like that to CISL, and we and they 
would put it into the "development" version of the compiler within a few 
days.

My favorite story about doing that was that we had a debate about using 
static links or a display to manage up-level references in the Ada/SIL 
compiler.  We had a ferocious call-graph analyzer that merged stack 
frames whenever possible, and of course, all variables in library 
packages (and in the main program if it was not called recursively) were 
handled directly.  So we put in a trigger that added the static pointers 
if necessary, and sent e-mail to a list of compiler people.  Almost two 
years later, I finally got one of those messages--for an ACVC test.

We decided we would take that code out and use a dynamic stack walk if 
it was really necessary.  I don't think that change ever happened--it 
was of course, very low priority.

-- 

                                           Robert I. Eachus

"The terrorist enemy holds no territory, defends no population, is 
unconstrained by rules of warfare, and respects no law of morality. Such 
an enemy cannot be deterred, contained, appeased or negotiated with. It 
can only be destroyed--and that, ladies and gentlemen, is the business 
at hand."  -- Dick Cheney

Re: Hierarchical States Machines

[Randy Brukardt]

"Martin Krischik" <krischik@users.sourceforge.net> wrote in message
news:1649831.NvcpCQcTn1@linux1.krischik.com...
...
> Is there any deepter reason why you did not use:
>
>      loop
>         <<continue>>
>          <bunch of conditions>
>              <Do some processing>
>              if <unusual condition> then
>                    <Handle unusual case>
>                    goto Continue;
>             end if;
>         <bunch of else conditions>
>         end if;
>         <normal processing>;
>     end loop;
>
> i.E. A style guide or a RM I am not aware of?

I usually want the loop head to be re-executed, so that if it gets modified
during maintenance (to a for-loop or a while-loop), it won't be necessary to
restructure the code. Besides, in most of these cases, the loop actually is
a while-loop, walking a list or other data structure.

                 Randy.