what should

println(4-2018.toString().length)

print in your opinion?

But that is not a unary-minus anymore though.

It seems normal for a method call to have priority over an operator call. unaryMinus is an operator, right?

And yes, I checked EXACTLY that on Ruby when it prints:

puts 2018.to_s.length; # 4
puts -2018.to_s.length; # 5
puts 4-2018.to_s.length; # 0
x = 2018;
puts -x.to_s.length; # -4

I'd like to check other languages as well, but I don't know which ones allow to call methods on integer literals (java does not)

unaryMinus is always an operator in kotlin

Some languages do this exception (notably Java), others do not (like Kotlin) and prefer to keep simpler parsing rule. I’m not sure there is one right answer.

yes. I know, grammar says so

Consider the parser just looking at the following string:

-2018

. How should it parse it? Is the minus sign supposed to be part of the literal, or an operation on the lone number? There is no good way for the parser to make that choice and resolve the ambiguity, other than eliminating the possibility of using it as part of the literal, since it needs to recognize similar things as an operator in other contexts

It is possible (Java does it), but then

-

operator becomes this weird exception. Compare:

4-2018.toString().length
4 - 2018.toString().length
- 2018.toString().length
-2018.toString().length

In Kotlin all of that is consistent in behavior. In Java…. it is not an issue, since you cannot do

4.toString()

and you cannot have extensions on

Int

in Java either. So it is logical for Java to do it the way Java does.