Consider this Kotlin code:
var parent: T? = null
get() = if (isParent) this as T else field
set(value) { field = if (value == null) null else value.parent }
val isParent: Boolean
get() = parent == null
var description = ""
get() = if (isParent) field else parent!!.description
set(value) { if (isParent) field = value else parent!!.description = value }
Assume that isParent
returns true
if this instance is a parent
instance. If not getParent()
will return the parent instance. In Java you are allowed to access directly field of a different instance of same class like this:
String getDescription() { return getParent().description; }
void setDescription(String value) { getParent().description = value; }
(I am not saying that is a best thing to do, I simplified it for demostration). Comparing to Java, it would be nice to be able to do following:
var description = ""
get() = parent.field
set(value) { parent.field = value }
However this does not work and unfortunately it makes the code less readable. Especially if you have a lot of such variables, which are bound to this parent
.
A backing field of a property can only be accessed from a getter or setter of that property, and only for the instance on which the getter or setter has been invoked. If you need to provide multiple ways to access an attribute of a class, you need to define two distinct properties, one of which has a backing field to store the data and another has a getter and setter referring to the first property.
class Foo {
var parent: Foo? = null
val parentOrSelf: Foo get() = parent ?: this
private var _description: String? = null
var description = ""
get() = parentOrSelf._description
set(value) { parentOrSelf._description = value }
}
Collected from the Internet
Please contact [email protected] to delete if infringement.
Comments