CODEWITHSUNDEEP
scalastream
freedev
freedev

Reputation: 30027

Scala stream filter behaviour

I'm playing with scala streams. Here is the stream of all integers starting from a given number.
I have added a println to trace every call of from function. 

def from(n: Int): Stream[Int] = n #:: from({ println(n); n + 1 })

val nats = from(0)         //> nats  : Stream[Int] =  Stream(0, ?)

nats.take(4).toList        //> 0
                           //| 1
                           //| 2
                           //| 3
                           //| res0: List[Int] = List(0, 1, 2, 3, 4)

As expected, this is the output of my scala worksheet. Then I have created the stream of all prime numbers.

def sieve(s: Stream[Int]): Stream[Int] = {
  s.head #:: sieve(s.tail.filter({ println( "---" ); _ % s.head != 0 }))
}                          //> sieve: (s: Stream[Int])Stream[Int]

val primes = sieve(from(2))//> primes  : Stream[Int] = Stream(2, ?)

primes.take(4).toList      //> 2
                           //| ---
                           //| 3
                           //| 4
                           //| ---
                           //| 5
                           //| 6
                           //| ---
                           //| res1: List[Int] = List(2, 3, 5, 7)

Here comes the question. I have made what in my opinion should be a little change, adding the x parameter instead of _ placeholder. Surprisingly the output is quite different:

def sieve(s: Stream[Int]): Stream[Int] = {
  s.head #:: sieve(s.tail.filter(x => { println("---"); (x % s.head) != 0 }))
}                          //> sieve: (s: Stream[Int])Stream[Int]

val primes = sieve(from(2))//> primes  : Stream[Int] = Stream(2, ?)

primes.take(4).toList      //> 2
                           //| ---
                           //| 3
                           //| ---
                           //| 4
                           //| ---
                           //| ---
                           //| 5
                           //| ---
                           //| 6
                           //| ---
                           //| ---
                           //| ---
                           //| res1: List[Int] = List(2, 3, 5, 7)

I don't understand why all these repetitions. What's wrong in using the explicit parameter?

Upvotes: 2

Views: 3120

Answers (2)

Alexey Romanov
Alexey Romanov

Reputation: 170713

The difference is because { println( "---" ); _ % s.head != 0 } is short for { println( "---" ); x => x % s.head != 0 } rather than { x => println( "---" ); x % s.head != 0 }.

In the first case you first call println and then return the function, so --- is printed once per filter call; in the second it's once per filter call per element of the stream being filtered (and because sieve is recursive and filters the stream again, you end up with multiple --- for each element of the output).

Upvotes: 3

pedrofurla
pedrofurla

Reputation: 12783

Let's compare the argument to sieve in both cases, have a close look:

s.tail.filter(     { println( "---" );       _ % s.head  != 0  })
s.tail.filter(x => { println( "--- " + x ); (x % s.head) != 0 })

Do you see the difference in the println calls?

Now the outputs

First:         Second:
//> 2          //> 2
//| ---        //| --- 3

The crucial difference:

println( "---"      )
println( "--- " + x )

In the 2nd case you are appending " "+x right after the dashes and that's the exact difference between the cases.

Upvotes: 0

Related Questions