Re: Bring Out Your New Perl Code

[zby]

What does the 'state' declaration really do here? How does it make $cash_in_store different from a normal closured variable?

Update: I analyzed it a bit - and I see it now - if $cash_in_store was not declared state then it would not be shared between $drower->[0], $drower->[1], $drower->[2] and $drower->[3]. But of course as a closured variable it still would work for individual drawers.

[grinder]

It's the exact difference between the following two code snippets:

  BEGIN {
    my $var = 10;
    sub next {
      return ++$var;
    }
  }

And the following:

  sub next {
    state $var = 10;
    return ++$var;
  }

Now let's have a show of hands, how many people prefer the First Way To Do It? ... hmm, I see no hands? I thought as much :)

Tip o' the hat to Roy Johnson for pointing out that it was a BEGIN block I wanted, not an lexical scope with a label BEGIN:. And you do need a BEGIN block, not a bare block, otherwise you can run into grief with uninitialised values (which is kind of the whole point).

[sgt]

Actually you should compare using a bare block (no BEGIN) which is the standard (older) way to have "state" variables.

I like 'state' but the old way is still useful if you need a pair of cooperating subs (or more):

% steph@ape (/home/stephan) %
% cat state_old.px
#!/usr/bin/perl+

{
  my $var = 10; 
  sub next1 { ++$var; }
  sub next1_odd {
      if ($var % 2 == 0) {
          ++$var;
      }
      else {
          ++$var; 
          ++$var;
      }
  }
}

$\ = qq{\n};
print next1() for 1..5;
print next1_odd();

% steph@ape(/home/stephan) %
% perl+ -w state_old.px
11
12
13
14
15
17

cheers --stephan

[pfaut]

It looks like it acts more like a static variable in a C routine. There is no closure in Limbic~Region's code yet the variable looks like it is expected to retain its value between calls.