File > Settings > Language & Frameworks > Kotlin > Enable K2 mode
It is necessary to restart the IDE to do the next step.
Popup the quick search with shift + shift
and type Registry...
, then search for:
kotlin.k2.only.bundled.compiler.plugins.enabled
And uncheck it.
Apply the plugin in the build.gradle.kts
or build.gradle
:
plugins {
id("com.javiersc.kotlin.kopy") version "$kopyVersion"
}
As the plugin uses under the hood the Atomicfu library, the performance can be improved by applying the Atomicfu plugin, it is important to apply the Kopy plugin before the Atomicfu one.
plugins {
id("com.javiersc.kotlin.kopy") version "$kopyVersion"
id("org.jetbrains.kotlin.plugin.atomicfu") version "$kotlinVersion"
}
The extension kopy
is available to configure the plugin:
The visibility
option allows changing the visibility of the copy
and invoke
functions. The
default value is Auto
, which uses the same visibility the primary constructor has, as the original
copy
function does after Kotlin 2.0.20.
Possible values:
KopyVisibility.Auto
(Default): The visibility of the primary constructor is used.KopyVisibility.Public
: The visibility of the generated functions will bepublic
.KopyVisibility.Internal
: The visibility of the generated functions will beinternal
.KopyVisibility.Protected
: The visibility of the generated functions will beprotected
.KopyVisibility.Private
: The visibility of the generated functions will beprivate
.
It is possible to have a more restrictive Kopy copy
and invoke
functions than the original one,
for example by providing the KopyVisiblity.Private
and the primary constructor being public
or
internal
. The original copy
function would be public
or internal
respectively, and the Kopy
functions would be private
.
Note
If the primary constructor visibility is more restrictive than the specified visibility, the primary constructor one is used.
import com.javiersc.kotlin.kopy.args.KopyVisibility
plugins {
id("com.javiersc.kotlin.kopy") version "$version"
}
kopy {
visibility = KopyVisibility.Private
}
The functions
option allows to decide which functions will be generated.
Possible values:
KopyFunctions.All
(default): Both,copy
andinvoke
functions, will be generated.KopyFunctions.Copy
: Only thecopy
function will be generated.KopyFunctions.Invoke
: Only theinvoke
function will be generated.
import com.javiersc.kotlin.kopy.args.KopyFunctions
plugins {
id("com.javiersc.kotlin.kopy") version "$version"
}
kopy {
functions = KopyFunctions.All
}
import com.javiersc.kotlin.kopy.Kopy
fun main() {
val house = House(
squareMeters = 100,
kitchen = Kitchen(
cat = Cat(
name = "Garfield",
age = 5,
numbers = listOf(1, 2, 3),
),
squareMeters = 10,
),
)
val house2: House = house.copy {
squareMeters = 200
kitchen.cat.name = "Felix"
kitchen.cat.age = 7
kitchen.cat.numbers.updateEach { it + 1 }
kitchen.squareMeters = 20
}
// House(squareMeters=200, kitchen=Kitchen(cat=Cat(name=Felix, age=7, numbers=[2, 3, 4]), squareMeters=20))
println(house2)
}
@Kopy
data class House(val squareMeters: Int, val kitchen: Kitchen)
@Kopy
data class Kitchen(val cat: Cat, val squareMeters: Int)
@Kopy
data class Cat(val name: String, val age: Int, val numbers: List<Int>)
copy
and invoke
create a new instance of the data class with the content specified. There is no
difference between both functions.
set
and =
do the same, assigning a value.
val house2: House = house.copy {
kitchen.cat.name = "Felix"
}
val house3: House = house.copy {
kitchen.cat.name.set("Felix")
}
update
is a lambda which allows updating the value of the property while having access to the
current value.
val house2: House = house.copy {
kitchen.cat.name.update { name -> "$name Jr." }
}
updateEach
is a lambda that allows updating the values of an Iterable
while having access to the
current value of each element.
val house2: House = house.copy {
kitchen.cat.numbers.updateEach { it + 1 }
}
Kopy is compatible with Kotlin Multiplatform projects if it is used exclusively on projects that applies the plugin.
Calling copy
or invoke
functions from Swift, or anything that is not Kotlin with the Kopy plugin
applied, will not work.
Kopy uses the K2 compiler plugin whereas KopyKat uses KSP to generate code, as a consequence there are different advantages and disadvantages.
- Kopy's code generation is faster than KopyKat's
- Kopy does not need to run any Gradle task to get feedback on the IDE:
- Autocompletion shows the
copy
andinvoke
functions instantly after annotating a data class with@Kopy
annotation. - Removing the
@Kopy
annotation instantly removes thecopy
andinvoke
functions without running a Gradle task or manually deleting thebuild
directory of a project. - As it is not necessary to assemble/build the project, the feedback loop is faster.
- Autocompletion shows the
- Build time should be better too (not tested).
- Kopy only adds 5 or 6 functions/properties to each data class, whereas KopyKat needs to generate builders and the functions/properties match the number of properties in the data class. In the future, the number of properties Kopy adds to each data class will be reduced to 1 or 2.
Similar to Kotlin Power Assert, this plugin works on
the call site, so it modifies the body of the copy
or invoke
lambdas. That means the plugin must
be applied to get it working, so it is not a good idea to use it in a library or an SDK as it will
force the users to apply the plugin. KopyKat generates all
the builders, so it does not have this limitation.
An error is shown in the IDE or at compilation time if the plugin is not applied and the copy
or
invoke
function from a class in another module is called, but it will not appear if the consumer
is a Java application/library.
The plugin transforms the lambda into what a developer would do manually with copy
functions,
that means the copy
or invoke
lambda can only work if the plugin is applied to the project it is
being called.
If the plugin is not applied, the copy
and invoke
function calls will be marked as
errors. Don't suppress them without applying the plugin as they will not work.
It is not necessary to suppress them manually, the Gradle plugin will suppress them automatically.
There is no reflection or mutability, the class will have some new functions and properties added.
The number is limited to 6 independently of the number of properties the data class has:
copy
functioninvoke
function_atomic
propertyset
functionupdate
functionupdateEach
function
When the Context Parameters feature is available, the number of properties and methods added to the data class will be reduced to only 1 or 2:
copy
function [optional]invoke
function [optional]
A new CopyScope
will be created, and it will be used to store the rest of properties, and it
will be added as a context parameter to the copy
and invoke
lambdas:
data class House(val squareMeters: Int, val kitchen: Kitchen) {
fun copy(block: CopyScope.() -> Unit): T = ...
operator fun invoke(block: CopyScope.() -> Unit): T = ...
}