Though I do believe you've reached par for this course.

Please don't produce this kind of code except in fun.

But since you ask, the trick is recursion. Here it is broken out:

sub r {
  my $s;
  $s .= chop for @_ = @_;
  $s . 0 ?
    ( &r , $s ) :
    ()
}
[download]

and now with better variable names, fewer implicit variables, rewritten loops, that kind of thing:

sub recursive_s2a {
  my $str_last;
  foreach my $str (@_ = @_) {
    $str_last .= chop($str);
  }

  if ($str_last . 0) {
    return (recursive_s2a(@_) , $str_last );
  }
  else {
    return;
  }
}
[download]

And now you see that the function creates the last string in the list to produce, tests whether that was the empty string, if it was it returns nothing, if it wasn't it then proceeds by recursion to generate the first strings, and adds the string it produced to that list.

sub r {my$s;$s.=chop for@_=@_;$s.0?(&r,$s):()}
[download]

  sub r {my$s;$s.=chop for@_;$s?(&r,$s):()}

        #0        1         2         3  3
        #123456789012345678901234567890123

  # jeff,john,mary
  # jjm,eoa,fhr,fny
[download]

I suppose I am missing something, but I dont see what.
Why are you doing @_=@_ and $s.0??
Is this an issue for an earlier version of perl? As far as I understand these are redundant ops. When does $s.0 evalute in a boolean sense different to $s? And what is the point of assigning @_ to itself?
I am confused.

OTOH it does mean that you can claim 5 less chars, for a total of 33.
:-)
Or am I being a lout?

Yves

In Perl 5.6.0 if you passed in literal strings, you got an error about, "Modification of a read-only value attempted". Recopying @_ fixed that.

And the time when $s and $s.0 are different in boolean context is when $s is "0". So when you remove those two characters, you get a bug.

UPDATE
An example that shows the bug:

print map "$_\n", r("0a1", "b0", "0");
[download]

Hi Tilly.

Ok so on that input it doesnt work.

I guess now I'm being a pain, but japhy's spec *did* say make a function that takes any number of equal-length strings

However I do believe in defensive programming, so I can see why you did that.

Very interesting, and food for thought

Yves

$\="\n"; $,=",";
print r("0000", "0000", "0000");
#outputs
#000,000,000,000
[download]

Hmm, and I reran it with literal params (I was using an @array) and I got that warnings as well. Interesting. OST. It follows from the fact @_ alises not copies, but still somehow surprising.

And it is also very interesting to learn that '00' evals to true. Definitely *not* what I expected. I was thinking that it would fall under the evals to 0 rule, but now I see not. Thanx.

On the other hand I tested a bunch of cases with 0's in them and I couldnt see how it affected the solution?

Yves