C: f>0 vs Perl: $f>0?

This C code displays Fibonacci numbers:

#inclu开发者_JAVA技巧de <stdio.h>
int main(){
   for (long long int t, i=1, p=1, f=2; f>0 ; i++, t=f, f+=p, p=t)
      printf("%lli: %lli\n", i, f); 
   }

As intended it stops without having to hard code the number of iterations.

The code in Perl is attempting to use the same technique:

my $t;
my $p=1;
my $f=2;
for ($i=1; $f>0; $t=$f, $f+=$p, $p=$t, $i++){
   printf("%i: %i\n", $i, $f);
   if ($i>50){exit;}
   }

But it does not stop as expected, at the point the value overflows at 45 rather than at 51 when the extra check kicks in.

44: 1836311903
45: -1323752223
46: -1
47: -1
48: -1
49: -1
50: -1
51: -1

What does Perl do differently that causes this?

---

Perl uses arbitrary types, switching between unsigned and signed integers and floating point as needed. printf, however, is forcibly converting the arbitrary type to an integer. Try using %s instead:

my $t;
my $p=1;
my $f=2;
for ($i=1; $f>0; $t=$f, $f+=$p, $p=$t, $i++){
   printf("%i: %s\n", $i, $f);
   if ($i>50){exit;}
}

The large negative number is in fact stored in perl as an unsigned integer; the -1's are too large floating point numbers being truncated to the maximum unsigned integer but then treated as signed by %i.

---

Your C signed integer code overflows from a large number to a negative number and stops. Perl's use of floating pt math lets it go through more iterations with reduced precision, and I don't think it'll wrap around when the number gets really big.

---

Your c code is actually exiting because of poor programming practice. You're relying on f being a long long int, which has an upper limit of 9.22337204 × 10^18. When you loop around to the 91st iteration f becomes 1.22001604 × 10^19, which overflows and becomes negative.

The Perl code continues on merrily but, as you should notice, eventually you'll print out a sequence of -1s, which should be a clue that your code is incorrect.

If you didn't have any limits on integer size your c code would go on forever, which is probably not what you want. For infinite sequences like the Fibonacci sequence you want to have a halting condition.

---

Possibly the Perl code defaults to floating-point arithmetic. That seems to be what this indicates. Try throwing a use integer; in there.

---

You are using 32-bit integers in Perl and larger integers in C. 32-bits are not enough to store those numbers. You could rebuild your Perl to use 64-bit integers, or you could let Perl use floating pointer numbers (53-bit integers with no loss) as ysth suggested.