Kotlin: fun vs val
To add to other answers, these are from the book Java to Kotlin, chapter 11 Methods to Properties:
Example 1
Suppose we want to add age to this class:
data class Person(val dateOfBirth: LocalDate)
We can compute age easily (ignoring time zones) from the dateOfBirth
property. But this doesn’t depend only on that property; it also depends on when we call it.
Unlikely though it is, fred.age == fred.age
can return false
.
Age is an action; its result depends on when it is called. Properties
should be calculations, timeless and dependent only on their inputs, in this case the dateOfBirth
property.
Hence, age()
should be a function, not a property:
data class Person(val dateOfBirth: LocalDate) {
fun age() = Period.between(dateOfBirth, LocalDate.now()).years
}
Example 2
What if we want a cryptographic hash of all the other properties of the object?
This is a calculation (for immutable objects), but if it is expensive to compute, it should be a method hash()
not a property hash
. We might even want to hint at the cost of the method in its name:
data class PersonWithProperties(
val givenName: String,
val familyName: String,
val dateOfBirth: LocalDate
) {
fun computeHash(): ByteArray =
someSlowHashOf(givenName, familyName, dateOfBirth.toString())
}
The official Kotlin Coding Conventions defines in section Functions vs Properties the following:
In some cases functions with no arguments might be interchangeable with read-only properties. Although the semantics are similar, there are some stylistic conventions on when to prefer one to another.
Prefer a property over a function when the underlying algorithm:
- does not throw
- is cheap to calculate (or caсhed on the first run)
- returns the same result over invocations if the object state hasn't changed
So, I would use in the above example a val
for isWhite
, since it does not throw, the string comparison is cheap to calculate and the color
of the Car
can't change, as the Car.color
is itself defined as val
.
Compiled difference
Note that the JVM bytecode of the get()
block will get compiled to the exact same code as the function would have. So, both approaches are the same regarding the compiled bytecode and there is no performance difference.