Swift: nil coalescing operator `??` measuring by mistake

[kijuky]

I feel that ?? is being measured incorrectly in some cases.

Case 1

func f0() {
  print(optionalValue ?? "nil")
}

I think ccn is 2. The reason is that it is equivalent to the following code, and one control code (if) is included.

func f0() {
  if let value = optionalValue {
    print(value)
  } else {
    print("nil")
  }
}

Case 2

func f0() {
  print(optionalValue?.value ?? "nil")
}

I think ccn is 2. The reason is that it is equivalent to the following code, and one control code (if) is included.

func f0() {
  if let value = optionalValue {
    print(value.value)
  } else {
    print("nil")
  }
}

Case 3

func f0() {
  print(optionalValue?.optionalValue?.value ?? "nil")
}

I think ccn is 3. The reason is that it is equivalent to the following code, and two control code (if) is included.

func f0() {
  if let value = optionalValue {
    if let value2 = value.optionalValue {
      print(value2.value)
    } else {
      print("nil")
    }
  } else {
    print("nil")
  }
}

Remark

• Similar code does not count ccn when using swiftlint. So my way of thinking may not be correct...
• According to the lizard policy in the [README](), this may be a trivial matter.
  • I use ccn as a reference for the number of test codes, so I want to measure the number of branches as accurately as possible.
• Once this issue's policy has been decided, I would like to contribute to the solution of this issue.

[terryyin]

I would count the Case 1 as ccn 2, too.

But in the case of 2 and 3, I think it will be quite hard for Lizard to identify the redundancy of ~?.value ?? ~ in a smart way. Unless we just ignore the sequence ?.value ???

[terryyin]

or just count ?.value ?? as +1 ccn only.

[kijuky]

Thank you for your comment!

I think implementing case 2 and case 3 with lizard would be a big fix.

So I want to confirm lizard's policy. Should it be measured accurately even if the processing becomes complicated to some extent? (However, I believe that the only change will be swift.py.) Is this case accurate to measure?

[kijuky]

I found a case where the lizard parser had limitations...

Case 2-1

func f0() {
  print(optionalValue?.optionalValue ?? "nil")
}

I think ccn is 3. The reason is that it is equivalent to the following code, and two control code (if) is included.

func f0() {
  if let value = optionalValue {
    if let value2 = value.optionalValue {
      print(value2)
    } else {
      print("nil")
    }
  } else {
    print("nil")
  }
}

Comparing Case 2 and Case 2-1, CCN differs depending on whether the last value is optional or not for expressions including unwrapping. However, lizard can not detect this.

[kijuky]

I believe that the consent for Case 1 has been obtained. So I want to add ?? to _conditions.

[kijuky]

I'm sorry. I was misunderstood.

In Swift, CCN does not increase because ? And ?? are method calls.

This issue is closed.