Re^2: reverse a string in place

[davido]

No built-in functions (abundant use of Perl's operators, though).

Yes, there's a spark of evil. But "they" always say that the easiest way is to divide a problem into smaller sub-problems:

sub rec_reverse {
  return $_[0] if $_[0] eq '';
  my( $char, $smaller_problem ) = $_[0] =~ m/\A(.)(.*)\z/s;
  return rec_reverse( $smaller_problem ) . $char;
}

print rec_reverse( "Hello\nworld." ), "\n";

I wouldn't call it sophisticated or elegant though, and if it were turned in as ones own work, it would probably merit some skepticism. Recursion is not an optimal approach here, since the number of recursive calls will match the length of the string, and each call added to the call-stack has a cost. Also, the regexp contortion as a means of eliminating any "built-in" functions (ie, substr) does nothing to dispel the rumor that Perl is difficult to read.

I find with recursion that it always helps me to think through it by thinking in terms of "problem" and "smaller_problem". I'm always tempted to write recursion like this:

sub rec_reverse {
  my $problem = shift;
  if( $problem eq '' or ! defined $problem ) {
    return $problem;    # Can't be made smaller. End case.
  }
  else {
    my( $char, $smaller_problem ) = $problem =~ m/\A(.)(.*)\z/s;
    return rec_reverse( $smaller_problem ) . $char;
  }
}

Which is the same thing but more explicit, and possibly more readable (and possibly less, since we've lost variable names that look like strings).