Directives: commit

                                         {my %variables}

start:          statement  /^\Z/         {$variables {'.'} = $item [1]}

statement:      variable '=' <commit> statement
                                         {$variables {$item [1]} = $item [4]}
              | expression

expression:     <leftop: term ('+' | '-') term>
                        {my $s = shift @{$item [1]};
                         while (@{$item [1]}) {
                             my ($op, $exp) = splice @{$item [1]}, 0, 2;
                             if ($op eq '+') {$s += $exp}
                             else            {$s -= $exp}
                         }
                         $s
                        }

term:           <leftop: factor m{([*/])} factor>
                        {my $p = shift @{$item [1]};
                         while (@{$item [1]}) {
                             my ($op, $term) = splice @{$item [1]}, 0, 2;
                             if ($op eq '*') {$p *= $term}
                             else            {$p /= $term}
                         }
                         $p
                        }

factor:         number
              | variable                 {$variables {$item [1]} ||=
                                                Math::BigInt -> new (0)}
              | '+' <commit> factor      {$item [3]}
              | '-' <commit> factor      {$item [3] * -1}
              | '(' <commit> statement ')'
                                         {$item [3]}

number:         /\d+/                    {Math::BigInt -> new ($item [1])}
variable:       /[.a-z]+/i

• The <commit> directive tells the parser that once this directive has been matched, none of the subsequent productions need to be tried if the current production fails.
• <commit> always matches, and its return value is 1.
• Benefit: less backtracking, hence more efficient.
• One can <uncommit> as well.