glenjamin/node-fib

Fast non-blocking fibonacci server

node-fib's Issues

Get a clue by actually understanding the OPs point

fewer dependencies

Perhaps some code like this with fewer dependencies (and less memory used) could work.
For 1000000 it returns Infinity, for 1000 it returns 2.686381002448534e+208 in 0.071s

function fibonacci(nfib, cb) {
  function fib(n1, n2, n, cb) {
    if(n > nfib) return cb(n2)
    process.nextTick(function() { fib(n2, n1 + n2, n + 1, cb) })
  }
  switch(nfib) {
    case 1: cb(0); break
    case 2: cb(1); break
    default: fib(0, 1, 3, cb)
  }
}

// usage example
// first arg is fibonacci number, 1-based
fibonacci(3, function(res) { console.log('fibonacci number is ', res) })

Cannot calculate 1,000,000th Fibonacci number

This crashes node-fib:

$ time curl localhost:3000/1000000
curl: (52) Empty reply from server

real    1m0.392s
user 0m0.006s
sys 0m0.004s

It appears it ran out of memory:

$ node app.js 
FATAL ERROR: CALL_AND_RETRY_2 Allocation failed - process out of memory

Should use closed-form fibonacci function

It would be much more performant to calculate the fibonacci number like so:

function fib (n) {
  var a=1
    , b=0
  while (0 > n--) {
    var c = b
    b = b + a
    a = c
  }
  return b
}

Perhaps it might make sense to add this as a separate endpoint?

Memoize results in an LRU cache

For optimum performance, the server should use node-lru-cache to memoize the results of each calculation in a lru cache. (npm install lru-cache is a pretty well tested O(1) implementation.)

Missing LICENSE.txt?

...a permissive license, I hope?

Improve ten-fold by knowing math

O(log n)

phi = (1 + Math.sqrt(5)) / 2
fibonacci = (n) -> Math.floor(Math.pow(phi, n) / Math.sqrt(5) + 1/2)
fibonacci(10) # => 55