Mirror of forum.nim-lang.org

4131 :: is space around "==" the part of syntax?

• oyster (orginal) [2018-08-16T07:23:15+02:00] view original

  
  echo 1 == 0 # space in the beginning and end of ==
  
  echo 1==0 # no space
  Run

  can be compiled and run without warning/error

  however

  
  echo 1== 0
  Run

  can be compiled and run as expected, but there is a warning ` Warning: Number of spaces around '==' is not consistent [Spacing]`

  the most strange is that nim refuses to compile

  
  echo 1 ==0
  Run

  and says

  
  Error: type mismatch: got <int literal(0)>
  but expected one of:
  proc `==`(x, y: float): bool
  proc `==`(x, y: int64): bool
  proc `==`[T](x, y: ptr T): bool
  proc `==`(x, y: int): bool
  proc `==`(x, y: string): bool
  proc `==`[T: proc](x, y: T): bool
  proc `==`[T](x, y: ref T): bool
  proc `==`[T: SomeUnsignedInt](x, y: T): bool
  proc `==`(x, y: bool): bool
  proc `==`(x, y: char): bool
  proc `==`[T](x, y: seq[T]): bool
  proc `==`[T: tuple |
      object](x, y: T): bool
  proc `==`[Enum: enum](x, y: Enum): bool
  proc `==`(x, y: float32): bool
  proc `==`(x, y: int8): bool
  proc `==`(x, y: int32): bool
  proc `==`(x, y: int16): bool
  proc `==`(x, y: cstring): bool
  proc `==`[T](x, y: set[T]): bool
  proc `==`(x, y: pointer): bool
  proc `==`[I, T](x, y: array[I, T]): bool
  
  expression: ==0
  Run

---

• Araq (orginal) [2018-08-16T08:37:59+02:00] view original

  No, what happens is that with your way of spacing it the == is interpreted as a unary operator and the error message says that too, only pass an integer literal to the == in the (sub)expression ==0.

---

• mratsim (orginal) [2018-08-16T08:42:30+02:00] view original

  The warning was (controversially) added so that people don't inadvertently use prefix notation when infix was wanted like 1 ==2.

  1 ==2 is parsed as (1, (==`(2)), i.e. `== is prefix while 1 == 2 is parsed as (==, 1, 2) so == is infix.

  This is especially important for the operator with both prefix and infix forms like - and &.

  By mistake you could use: a &b instead of a & b. More details in #7685.

---