Fibonacci numbers

Here it is if you've ever wanted to have a class that creates them, and we all do, I just know it :-) This is a quick class I wrote while being bored, if anyone wants to make comments I'd welcome them. :-)

Code:

---

#include <iostream>
#include <cstdlib>
#include <cmath>

using namespace std;

class Fibonacci {
  unsigned long *cache;
  int cache_max;
public:
  explicit Fibonacci(int max = 46);
  ~Fibonacci();

  int max();
  unsigned long get(int);
  unsigned long operator[](int);
};

Fibonacci::Fibonacci(int max)
  : cache_max(max + 1)
{
  cache = new(nothrow) unsigned long[cache_max];

  if (cache == 0)
  {
    cerr<<"No memory"<<endl;
    exit(1);
  }

  for (int i = 0; i < cache_max; i++)
  {
    cache[i] = 0;
  }
}

Fibonacci::~Fibonacci()
{
  delete cache;
}

int Fibonacci::max()
{
  return cache_max;
}

unsigned long Fibonacci::get(int n)
{
  if (n > cache_max || n < 0)
  {
    cerr<<"Subscript out of range"<<endl;
    return 0;
  }
  else if (cache[n] == 0)
  {
    double s = sqrt(5.0);
    double x = 0.5 * s;

    // Deep magic
    cache[n] = ((pow(0.5 + x, n)) - (pow(0.5 - x, n))) / s;
  }

  return cache[n];
}

unsigned long Fibonacci::operator[](int sub)
{
  return get(sub);
}

int main()
{
  Fibonacci fib;

  for (int i = 0; i < fib.max(); i++)
  {
    cout<< fib[i] <<endl;
  }
}

---

Wow, when you're bored, you're REALLY bored! :D

Me likes by the way!

>>Wow, when you're bored, you're REALLY bored!
Tell me about it, I actually wrote it in Perl first, then ported it to C++. Next I'll add operator overloading to the Perl script. :-)

Code:

---

#!usr/bin/perl -w

use strict;

package Fibonacci;

sub make {
  my $self = shift;
  my $ref = {cache_max => 47, cache => []};

  bless $ref, $self;
}

sub get {
  my ($self, $sub) = @_;

  if ($sub < 0 || $sub > $self->{cache_max}) {
    print STDERR "Subscript is out of range\n";
  }

  return $self->{cache}->[$sub] if defined $self->{cache}->[$sub];

  my $s = sqrt 5.0;
  my $x = 0.5 * sqrt 5.0;

  # Deep magic
  $self->{cache}->[$sub] =
    int((((0.5 + $x) ** $sub) - ((0.5 - $x) ** $sub)) / $s);
}

sub max {
  my $self = shift;

  $self->{cache_max};
}

package main;

my $fib = Fibonacci->make;

for (my $i = 0; $i < $fib->max; $i++) {
  print $fib->get($i), "\n";
}

---

Minor bug...delete cache should be delete [] cache...but yeah, you were right, I've always wanted the code for that one. :D

One more comment, too. IMO you needn't bounds check on the array access's because of the speed tradeoff of the two comparisons. Most code will/should mind it's own bounds...and if it does then not only does your code do bounds checking, so does their's! :-/

- cheers.

>>Minor bug...delete cache should be delete [] cache
Yikes, I can't believe I missed that. Thanks for your sharp eyes. That's a major bug by the way, it's undefined behavior. :-)

>>IMO you needn't bounds check on the array access's because of the speed tradeoff of the two comparisons.
I like to be neighborly and protect people against themselves when I can :-) But a middle ground is probably better, give the user a choice of a range checked get() for more compact code, or non checked get() for speedyness, like the [] and at() methods of vector.