Request for critics

Request for critics

[Maciej Sobczak]

For the purpose of my learning, please criticise the following code:

with Ada.Text_IO;
use Ada.Text_IO;

procedure Automata is

    Size : constant := 66;
    Max_Steps : constant := 31;

    type Cell_State is (Active, Inactive);
    type Row_Of_Cells is array (1 .. Size) of Cell_State;

    type Cell_Triplet is array (-1 .. 1) of Cell_State;

    Automata : Row_Of_Cells := (Size / 2 => Active, others => Inactive);

    procedure Show(Row : in Row_Of_Cells) is
    begin
       for I in Row_Of_Cells'Range loop
          case Row(I) is
             when Active   => Put('*');
             when Inactive => Put(' ');
          end case;
       end loop;
       New_Line;
    end Show;

    function New_State(Neighbourhood : in Cell_Triplet)
                       return Cell_State is
       Left  : Cell_State := Neighbourhood(-1);
       Right : Cell_State := Neighbourhood( 1);
    begin
       if (Left = Active and Right = Inactive)
         or (Left = Inactive and Right = Active) then
          return Active;
       else
          return Inactive;
       end if;
    end New_State;

begin
    Show(Automata);
    for Step in 1 .. Max_Steps loop
       declare
          New_Automata : Row_Of_Cells := (others => Inactive);
       begin
          for Cell_Index in
            Row_Of_Cells'First + 1 .. Row_Of_Cells'Last - 1 loop

             New_Automata(Cell_Index) := New_State(Cell_Triplet(
                Automata(Cell_Index - 1 .. Cell_Index + 1)));

          end loop;
          Automata := New_Automata;
       end;
       Show(Automata);
    end loop;
end Automata;

<problem-description>
The above is a short simulation of one particular 1D cellular automata 
(the state of the whole automata is printed as a single line and the 
time progresses downwards on the printout), where the next state of each 
cell is a function of the current states in the immediate neighbourhood 
of the given cell. In other words, given ABCDEFGH as a state of the 
automata, the next state of cell D is a function of cells C, D and E. 
Taking two possible states, there are 256 possible rules. The above 
program simulates only one of them.
</problem-description>

The goal is the readability and proper expression of the problem.

How do you name loop control variables for array traversals? Is 
"Cell_Index" overly verbose and would be "I" just enough?

Would you drop the Cell_Triplet type and pass the neighbourhood as three 
separate parameters instead?

And so on. Pleasy be picky with every aspect you find important.

Any performance considerations for this code (except, of course, that 
Put() is called for each cell and might be replaced by whole-line 
printouts instaed)?


BTW: The C version is just a bit shorter:

#include <stdio.h>
int main(){char s[]=
"                                *                                 "
;char p,n;int st,i;puts(s);for(st=0;st!=31;++st){p=' 
';for(i=1;i!=65;++i){n=(s[i-1]^s[i+1])|0x20;s[i-1]=p;p=n;}s[i-1]=p;puts(s);}return 
0;}

;-)  (the last three characters are not part of the program)

-- 
Maciej Sobczak : http://www.msobczak.com/
Programming    : http://www.msobczak.com/prog/

Re: Request for critics

[Gautier]

Just an opinion:

       if (Left = Active and Right = Inactive)
         or (Left = Inactive and Right = Active) then

seems to me "over-explicit" and indeed less readable than:

       if Left /= Right then

_______________________________________________________________ 
Gautier         -- http://www.mysunrise.ch/users/gdm/index.htm 
Ada programming -- http://www.mysunrise.ch/users/gdm/gsoft.htm 

NB: For a direct answer, e-mail address on the Web site!

Re: Request for critics

[Jeffrey R. Carter]

Maciej Sobczak wrote:
> For the purpose of my learning, please criticise the following code:

Sure. I love to tear apart others' work :)

> with Ada.Text_IO;
> use Ada.Text_IO;

Given 2 calls to Put and 1 to New_Line, all in 1 procedure, a global use of 
Ada.Text_IO seems inappropriate;

> procedure Automata is
> 
>    Size : constant := 66;
>    Max_Steps : constant := 31;

This appears not to be a maximum (which implies it may not be reached), but a 
number, so I'd probably call it Num_Steps.

> 
>    type Cell_State is (Active, Inactive);
>    type Row_Of_Cells is array (1 .. Size) of Cell_State;
> 
>    type Cell_Triplet is array (-1 .. 1) of Cell_State;

Cell_Triplet, with its attendant type conversion, seems inappropriate. How about 
something like

type Cell_List is array (Positive range <>) of Cell_State;
subtype Row_Of_Cells is Cell_List (1 .. Size); -- Cell_Row?
subtype Cell_Triplet is Cell_List (1 .. 3);

You can then call New_State with a slice of a Row_Of_Cell of length 3.

>    Automata : Row_Of_Cells := (Size / 2 => Active, others => Inactive);

Personally, I'd put this after the subprograms, simply to avoid accidental use 
of it from within them. I also dislike the potential confusion of reusing the 
identifier Automata like this.

>    procedure Show(Row : in Row_Of_Cells) is

You can put your use clause here, or fully qualify the 3 calls.

>    begin
>       for I in Row_Of_Cells'Range loop

I prefer to base indices on objects rather than types:

for I in Row'range loop

I also like to name all loops, but I seem to be in the minority about that.

>          case Row(I) is
>             when Active   => Put('*');
>             when Inactive => Put(' ');
>          end case;
>       end loop;
>       New_Line;
>    end Show;
> 
>    function New_State(Neighbourhood : in Cell_Triplet)
>                       return Cell_State is
>       Left  : Cell_State := Neighbourhood(-1);
>       Right : Cell_State := Neighbourhood( 1);
>    begin
>       if (Left = Active and Right = Inactive)
>         or (Left = Inactive and Right = Active) then

This seems to be

if Left /= Right then

which also seems clearer.

>          return Active;
>       else
>          return Inactive;
>       end if;
>    end New_State;

Neighbourhood (0) is not referenced, so perhaps it shouldn't be passed, either.

> begin
>    Show(Automata);

I'd like to see a blank line here.

>    for Step in 1 .. Max_Steps loop
>       declare
>          New_Automata : Row_Of_Cells := (others => Inactive);
>       begin
>          for Cell_Index in
>            Row_Of_Cells'First + 1 .. Row_Of_Cells'Last - 1 loop
> 
>             New_Automata(Cell_Index) := New_State(Cell_Triplet(
>                Automata(Cell_Index - 1 .. Cell_Index + 1)));
> 
>          end loop;
>          Automata := New_Automata;
>       end;

And here.

>       Show(Automata);
>    end loop;
> end Automata;

> The goal is the readability and proper expression of the problem.
> 
> How do you name loop control variables for array traversals? Is 
> "Cell_Index" overly verbose and would be "I" just enough?

I don't have any problem with either, and would probably use Index.

> Would you drop the Cell_Triplet type and pass the neighbourhood as three 
> separate parameters instead?

I'd probably just pass Left and Right. But, consider making this a generic, with 
New_State as a generic parameter (and Size and Max_Steps as parameters), and 
think about what the parameter list for New_State should be then.

> BTW: The C version is just a bit shorter:
> 
> #include <stdio.h>
> int main(){char s[]=
> "                                *                                 "
> ;char p,n;int st,i;puts(s);for(st=0;st!=31;++st){p=' 
> ';for(i=1;i!=65;++i){n=(s[i-1]^s[i+1])|0x20;s[i-1]=p;p=n;}s[i-1]=p;puts(s);}return 
> 0;}
> 
> ;-)  (the last three characters are not part of the program)

Indeed, and as C lovers love to point out, shorter things are more readable (see 
my comment on your condition above), so this must be clearer, as well :)

Seriously, part of the reduction is the use of a string to represent the state; 
this could be incorporated in Ada as well, eliminating Show. That would make the 
Ada shorter, too. For this little example, it doesn't much matter, but in 
general modeling the problem directly by using appropriate user-defined types is 
better. Another part is incorporating the update algorithm in line, but using a 
function is clearer and allows for parameterization.

The formatting also helps reduce the size, but I would not recommend this 
formatting style (except for C :).

By the way, automata is the plural of automaton, but you seem to use it as a 
singular.

-- 
Jeff Carter
"Son of a window-dresser."
Monty Python & the Holy Grail
12

Re: Request for critics

[tmoran]

A few additional minor suggestions:
     Image : constant array(Cell_State) of Character
       := (Active => '*', Inactive => ' ');
then
     procedure Show(Row : in Row_Of_Cells) is
       use Ada.Text_IO;
     begin
        for I in Row'range loop
          Put(Image(Row(I)));
        end loop;
        New_Line;
     end Show;

and
     New_State : constant array(Cell_State, Cell_State) of Cell_State
       := (Active   => (Active=>Inactive, Inactive=>Active),
           Inactive => (Active=>Active, Inactive=>Inactive),
then
     New_Automata(Cell_Index) := New_State(Automata(Cell_Index - 1),
                                           Automata(Cell_Index + 1));

Re: Request for critics

[Steve]

I've been writing a lot of time critical code lately so my approach to 
coding may be a little out of the norm, but here is some code with the same 
functionality.  Columns should align nicely if you view them in a fixed 
pitch font.:

with Ada.Text_IO;
use Ada.Text_IO;

procedure Automata is

    Size      : constant := 66;
    Max_Steps : constant := 31;

    type Cell_State is (Active, Inactive);
    type Row_Of_Cells is array (1 .. Size) of Cell_State;

    type State_To_Char is array( Cell_State'Range ) of Character;
    State_Char : constant State_To_Char := ( Active => '*', Inactive => ' 
' );

    procedure Show(Row : in Row_Of_Cells) is
    begin
       for I in Row_Of_Cells'Range loop
          Put( State_Char( Row(I) ) );
       end loop;
       New_Line;
    end Show;

    type State_Map is array( Cell_State'Range, Cell_State'Range, 
Cell_State'Range ) of Cell_State;

    New_State : constant State_Map :=
       ( Active =>   ( Active =>   ( Active   => Inactive,     -- Active 
Active   Active   => Inactive
                                     Inactive => Active ),     -- Active 
Active   Inactive => Active
                       Inactive => ( Active   => Inactive,     -- Active 
Inactive Active   => Inactive
                                     Inactive => Active ) ),   -- Active 
Inactive Inactive => Active
         Inactive => ( Active =>   ( Active   => Active,       -- Inactive 
Active   Active   => Active
                                     Inactive => Inactive ),   -- Inactive 
Active   Inactive => Inactive
                       Inactive => ( Active   => Active,       -- Inactive 
Inactive Active   => Active
                                     Inactive => Inactive ) )  -- Inactive 
Inactive Inactive => Inactive
       );

begin
    declare
       Automata : Row_Of_Cells := (Size / 2 => Active, others => Inactive);
    begin
       Show(Automata);
       for Step in 1 .. Max_Steps loop
          declare
             New_Automata : Row_Of_Cells := (others => Inactive);
          begin
             for Cell_Index in
               Row_Of_Cells'First + 1 .. Row_Of_Cells'Last - 1 loop

                New_Automata(Cell_Index) := New_State(
                   Automata(Cell_Index - 1),
                   Automata(Cell_Index),
                   Automata(Cell_Index + 1));

             end loop;
             Automata := New_Automata;
          end;
          Show(Automata);
       end loop;
    end;
end Automata;


"Maciej Sobczak" <no.spam@no.spam.com> wrote in message 
news:dvr7a5$9o3$1@sunnews.cern.ch...
> For the purpose of my learning, please criticise the following code:
[snip]

> <problem-description>
> The above is a short simulation of one particular 1D cellular automata 
> (the state of the whole automata is printed as a single line and the time 
> progresses downwards on the printout), where the next state of each cell 
> is a function of the current states in the immediate neighbourhood of the 
> given cell. In other words, given ABCDEFGH as a state of the automata, the 
> next state of cell D is a function of cells C, D and E. Taking two 
> possible states, there are 256 possible rules. The above program simulates 
> only one of them.
> </problem-description>
>
> The goal is the readability and proper expression of the problem.
>
> How do you name loop control variables for array traversals? Is 
> "Cell_Index" overly verbose and would be "I" just enough?
>
In this example of code, IMHO "I" would be enough.
There is only one index involved and the loop is small.  Personally I tend 
to use a more expressive index name for all cases, and would probably use 
Cell_Index, Data_Index, Index, or something similar.

> Would you drop the Cell_Triplet type and pass the neighbourhood as three 
> separate parameters instead?
>

In this case I would rop Cell_Triplet since it doesn't add anything to the 
readability and introduces a new data type and required conversions.  In my 
experience, for the simple cases less is better.

> And so on. Pleasy be picky with every aspect you find important.
>
> Any performance considerations for this code (except, of course, that 
> Put() is called for each cell and might be replaced by whole-line 
> printouts instaed)?

The performance change you'll note in my modified implementation is the use 
of a lookup table for New_State instead of a function.  I think it is easier 
to interpret the mapping that takes place in the lookup table than analyzing 
the code.

You'll also note that I moved the definition of Automata into a declare 
block in the main procedure.  I strive to restrict the scope of variables as 
much as possible.

Normally I also avoid using variable names that are the same as package 
names.  It may be legal, but it may also be confusing.

In your original code the state mapping really only depended on the two 
neighbors, so the State_Map really only needs to be 2x2 rather than the 3x3 
in my example.  I chose to implement 3x3 since it would make it easier to 
define other mappings.

>
> BTW: The C version is just a bit shorter:
>
> #include <stdio.h>
> int main(){char s[]=
> "                                *                                 "
> ;char p,n;int st,i;puts(s);for(st=0;st!=31;++st){p=' 
> ';for(i=1;i!=65;++i){n=(s[i-1]^s[i+1])|0x20;s[i-1]=p;p=n;}s[i-1]=p;puts(s);}return 
> 0;}
>
> ;-)  (the last three characters are not part of the program)
>

You can generate line noise in Ada too.  But it is a bit harder to do, and 
after all that is not the objective.

I hope this helps,

Steve
(The Duck)

> -- 
> Maciej Sobczak : http://www.msobczak.com/
> Programming    : http://www.msobczak.com/prog/ 

Re: Request for critics

[Justin Gombos]

On 2006-03-22, Maciej Sobczak <no.spam@no.spam.com> wrote:
>
>     Automata : Row_Of_Cells := (Size / 2 => Active, others =>
>     Inactive);

One thing that hasn't been caught yet - Automata should be a constant:

  Automata : constant Row_Of_Cells := (Size / 2 => Active, others => Inactive);

because it never changes after its declaration.

-- 
PM instructions: do a C4esar Ciph3r on my address; retain punctuation.

Re: Request for critics

[Stephen Leake]

Justin Gombos <rpbkbq.xax.gld@uluv.kbq> writes:

> On 2006-03-22, Maciej Sobczak <no.spam@no.spam.com> wrote:
>>
>>     Automata : Row_Of_Cells := (Size / 2 => Active, others =>
>>     Inactive);
>
> One thing that hasn't been caught yet - Automata should be a constant:
>
>   Automata : constant Row_Of_Cells := (Size / 2 => Active, others => Inactive);
>
> because it never changes after its declaration.

If you turning on all warnings in GNAT (-gnatwa), it will tell you
this automatically.

-- 
-- Stephe

Re: Request for critics

[Jeffrey R. Carter]

Stephen Leake wrote:
> Justin Gombos <rpbkbq.xax.gld@uluv.kbq> writes:
> 
>> On 2006-03-22, Maciej Sobczak <no.spam@no.spam.com> wrote:
>>>     Automata : Row_Of_Cells := (Size / 2 => Active, others =>
>>>     Inactive);
>> One thing that hasn't been caught yet - Automata should be a constant:
>>
>>   Automata : constant Row_Of_Cells := (Size / 2 => Active, others => Inactive);
>>
>> because it never changes after its declaration.
> 
> If you turning on all warnings in GNAT (-gnatwa), it will tell you
> this automatically.

Except it won't in this case, because there is an assignment to Automata.

-- 
Jeff Carter
"I spun around, and there I was, face to face with a
six-year-old kid. Well, I just threw my guns down and
walked away. Little bastard shot me in the ass."
Blazing Saddles
40