Infix notation and with(...) does not work as I expected - kotlin

Consider the following scenario:
I have a class Test
class Test() {
infix fun say(msg: String) = println(msg)
}
and a main method
fun main(args: Array<String>) {
val test = Test()
test say "Hello World!" //Works
with(test) {
say "Goodbye World!" //Does not work
say("Hello again!") //Works
}
}
As you can see I'm testing out the infix notation. Considering with(...) allows you to work with the object passed as parameter in the with block
without having to access its members through the dot notation, I would expect the infix notation to work like I show in my example above.
Unfortunately this does not work, is there a reason why this does not work? Is it a bug or simply a limitation? Or perhaps I am not interpreting the with(...) function correctly?

Infix notation is about the syntax of the way it's used. It works with an object on the left and the parameter on the right.
When using with you no longer have an object token on the left, so the special syntax for infix notation no longer works. You have to fall back to the regular function notation.

This restriction is necessary for the parser to parse the code without conflicts with other syntax.

Related

How do I specify an ActionListener in Kotlin?

I want to add an ActionListener to a JButton in Kotlin. In Java, I would just write this:
JPanel makeButtonPanel() {
JPanel panel = new JPanel(new FlowLayout());
JButton dirButton = new JButton("Change directory");
dirButton.addActionListener(e -> chooseDirectory());
panel.add(dirButton)
return panel;
}
But it's not so simple in Kotlin. I first tried this:
private fun makeButtonPanel() : JPanel {
val panel = JPanel(FlowLayout())
val dirButton = JButton("Choose")
dirButton.addActionListener(e -> chooseDirectory()) // error message here
// ...
}
private fun chooseDirectory() { ... }
But I'm getting this error message:
Type Mismatch
Required: ((ActionEvent!) -> Unit)!
Found: KFunction1<ActionEvent, Unit>
I understand that the ! means that this is a java method with uncertain nullability, but that doesn't help me understand how to write it. All I want it to do is call the chooseDirectory() method. There must be a clean, simple way to do this, but I don't see it.
As you've discovered, you need to use braces ({ }).
This is because braces are a necessary part of defining a lambda in Kotlin.  (That differs from languages like Java and Scala, where the necessary part is the -> or => arrow.  That's because in Kotlin the arrow is optional if there are one or no parameters; if one, the it keyword is used.)
Without the braces, the code would call your chooseDirectory() function, and try to pass its result to addActionListener() — which obviously wouldn't work.
Braces are also sufficient: they're taken as defining a lambda unless you're giving the body of a function or method or an if/when branch.  (Again, this differs from most C/Java-like languages.  In Kotlin, if you just want a block scope, you have to use a construct such as run.)
As for the parentheses, they're optional here.  You could include them if you wanted:
dirButton.addActionListener({ chooseDirectory() })
But Kotlin has a convention that if a function's last parameter is a function, you can pass it after the parens:
dirButton.addActionListener(){ chooseDirectory() }
And if that would make the parens empty, then you can omit them entirely:
dirButton.addActionListener{ chooseDirectory() }
That's to allow functions that look like new language syntax.  For example, you may have met the with function:
with(someObject) {
itsProperty = someValue
}
That's just a perfectly ordinary function, defined in the standard library, and taking a function as its last parameter.  Similarly, repeat:
repeat(10) {
// Some code to be run 10 times…
}
There's one further thing worth mentioning here.  In Kotlin, lambdas are one way to define functions, which are first-class types and can be defined, passed around, and used just like other types.  This differs from Java, which has traditionally used interfaces for those purposes — often interfaces with a Single Abstract Method (‘SAM interfaces’) — and in which lambdas are little more than syntactic sugar for defining an anonymous implementation of such an interface.
As a special case, for interoperability, Kotlin allows a lambda to define an implementation of a Java SAM interface (or, since Kotlin 1.4, of a Kotlin fun interface), instead of a function.
ActionListener is a Java SAM interface, which is why you can use a lambda here.
Okay, I figured it out, and it was pretty simple. I just have to dispense with the parentheses and say
dirButton.addActionListener { chooseDirectory() }
I'm still not clear on when I should use braces instead of parentheses.

What's the advantage of creating a one line function that contains a string, that wouldn't be achievable by just assigning a string to a variable?

In Kotlin, I understand that a string can be assigned to a function directly, such as:
fun foo(): String = "Hello World"
But you can also assign a String to a variable directly as well:
var foobar: String = "Hello Word"
My question is, why would you ever create a function when you could just create a variable? I can't see the point in the existence of this functionality.
The interesting thing about
fun foo(): String = "Hello World"
is that it is using the expression syntax, and is equivalent to:
fun foo(): String {
return "Hello World"
}
While just returning a constant isn't very useful, using the expression syntax, avoiding the {block} and return statements allows much more concise function definitions in the case where the entire function can be expressed in one expression.
For example, if foo() was a method on a class, you could say hello with a property of that class:
class Hello(var what : String = "World") {
fun foo(): String = "Hello, $what!"
}
fun main() {
val hello = Hello()
println(hello.foo())
hello.what = "Universe"
println(hello.foo())
}
This prints:
Hello, World!
Hello, Universe!
This is more about when to prefer a function v/s a property.
Kotlin coding conventions has a section that describes this.
A property should be preferred over a function when-
the underlying algorithm does not throw
value is cheap to calculate (or caсhed on the first run)
returns the same result over invocations if the object state hasn't changed
In terms of an API use-case, in some cases exposing a function instead of a property might be a good idea as that gives you the scope to change the implementation of this API in future. What might be a hardcoded value today, could be replaced by code that computes the value in future.
It's simple, think about what the word coding means. Coding rules. Rules that are complicated get broken down in rules that are exactly one level below in abstraction, so that the program is as intelligible as possible.
Well, the function name is just one level above the expression. This is even more true in a language as expressive as kotlin where one line can easily be equivalent to several lines of Java.
If you are talking about strings or primitives exclusively then, yes, an attribute is a more natural choice than a function.

What do grave accents do/mean in Kotlin?

In as many expressions/definitions as possible please.
I'm writing a test function, where after the call fails, the function returns:
`this `fails with` "the state is propagated"`
(with the grave accents surrounding fails with ^ i don't know how to escape, sorry)
You want to use them when something is a Kotlin keyword (like Java's System.in) but you need to call it. Then you can do
System.`in`
instead to make it work.
You can also use this in variables, functions, classes and any other identifiers. There is a small paragraph on this topic on Kotlin's documentation.
Actually, it is more than that.
You can use any class, function, variable, or identifier whose name contains spaces or symbols with grave accents.
class `Class name with spaces` {
fun `method name with spaces, +, -`(`a parameter`: Int) {
val `variable?!` = `a parameter` + 1
println(`variable?!`.toString())
}
}
fun main(args: Array<String>) {
val instance = `Class name with spaces`()
instance.`method name with spaces, +, -`(100)
}
This is a compilable and working code:
This is often used in testing, in order to make the test method names self-explanatory.
class OperationsUnitTest {
#Test
fun `addition should be commutative`() {
...
}
}

final or val function parameter or in Kotlin?

Why does Kotlin removed the final or val function parameter which is very useful in Java?
fun say(val msg: String = "Hello World") {
msg = "Hello To Me" // would give an error here since msg is val
//or final
...
...
...
}
Kotlin function parameters are final. There is no val or final keyword because that's the default (and can't be changed).
After Kotlin M5.1 support of mutable parameters removed, In earlier versions that can be achieve using
fun foo(var x: Int) {
x = 5
}
According to Kotlin developers, main reasons of removing this feature are below -
The main reason is that this was confusing: people tend to think that this means passing a parameter by reference, which we do not support (it is costly at runtime).
Another source of confusion is primary constructors: “val” or “var” in a constructor declaration means something different from the same thing if a function declarations (namely, it creates a property).
Also, we all know that mutating parameters is no good style, so writing “val” or “var” infront of a parameter in a function, catch block of for-loop is no longer allowed.
Summary - All parameter values are val now. You have to introduce separate variable for re-initialising. Example -
fun say(val msg: String) {
var tempMsg = msg
if(yourConditionSatisfy) {
tempMsg = "Hello To Me"
}
}
And another reason is that val and var differ by only one letter. This can be very confusing. So for function parameters they removed the option completely. Thus eliminating the confusion in this one small area (yet keeping it everywhere else--go figure).
This decision was made to avoid fragile base class problem. It happens when a small change in base classes (superclasses) makes subclasses malfunction.

How to define a oneline function calling a nullable variable function

If I have something like:
fun showProgressView() = ultraRecyclerView?.showProgressBar()
it says that it returns Unit? and not Unit (edited)
-----EDIT-----
One way can be
fun showProgressView() = ultraRecyclerView?.showProgressBar() ?: Unit
but it looks not right for me.
Another way:
fun showProgressView() { ultraRecyclerView?.showProgressBar() }
But I cant find a way for android studio maintain that format.
If you use the short expression form of a function, the inferred result type of the expression determines the function return type. If that is a platform type from Java, it could be nullable. If it is a Kotlin type then it will know the correct nullability.
But since you use the safe operator ?. you are saying for sure it could be nullable. And if the result is null or Unit then that gives the inferred result type of Unit?
Which is odd, but is exactly what you are saying. Therefore, either use a normal function body with { .. } or give the function an explicit return type if possible.
fun showProgressView(): Unit { ultraRecyclerView?.showProgressBar() }
You can also erase the nullability, by creating an extension function on Unit?:
fun Unit?.void() = Unit
And use it whenever you want to fix the return type:
fun showProgressView() = ultraRecyclerView?.showProgressBar().void()
IntelliJ IDEA / Android Studio do not appear to have a setting to keep the style of a block body function on a single line. Even so, you can use run to get around this:
fun showProgressView() = run<Unit> { ultraRecyclerView?.showProgressBar() }
Normally you do not need to add explicit type arguments to run but in this case providing them gives you the desired method signature (return type of Unit and not Unit?).
Note: This answer is adapted from a comment I gave to why assignments are not statements.