scala: 'def foo = {1}' vs 'def foo {1}'

Question

what is going on in each of these forms of defining foo?:

scala> def foo = {1}
foo: Int

scala> foo
res2: Int = 1

But:

Answer 1

found this in http://anyall.org/scalacheat/:

//[bad!] 
def f(x: Int) { x*x } //hidden error: without = it's a Unit-returning proc; causes havoc 

Answer 2

See also this question and answer on SO:

In Scala if a method declaration does not have an equal sign before its body, the compiler infers that the result type will be Unit

Basically declaring a function with no = means that the function returns Unit and the compiler inserts a () for you at the end. A function which should return a non-Unit value must be declared with the = notation (although of course the compiler can infer the return-type from the expression's type).