This is a general question about callback functions, defined in Kotlin Native, called by C functions.
For argument's sake, let's say I'm trying to walk a directory in a filesystem recursively, using https://linux.die.net/man/3/nftw in Kotlin Native.
(I know there are other ways to do this, using other C functions, but that is not the core of this question.)
nftw() takes a function as a callback:
val directory = "//some/directory"
val callback = kotlinx.cinterop.staticCFunction {
file: CPointer<ByteVar>?,
stat: CPointer<stat>?,
typeFlag: Int,
b: CPointer<FTW>? ->
val fileName = file?.toKString()
println(fileName)
val result = 0
result
}
val depth = 10
val flags = 0
platform.posix.nftw(directory, callback, depth, flags)
This works for listing files via "println()", but as as soon as the lambda contains any captured value, I get the following compiler error:
"kotlinx.cinterop.staticCFunction must take an unbound, non-capturing function or lambda".
My question is: is there any recommended approach on how to
access any non-global state from such a callback?
I did come up with a nasty workaround using a global variable, so that's not what I'm looking for primarily. If there is a commonly accepted solution using #ThreadLocal or something else, please discuss.
For native callbacks in general (not specific to nftw). The C function should accept a void* userData parameter and pass it to the callback when it is called. This allows you to pass local data to the callback, instead of global data.
This is the case even in C/C++.
For this particular case (regardless of language) there isn't really a way to do this without some global data (or JIT but let's not think about it). So any workaround would have to be nasty.
Using #ThreadLocal is a "reasonable nasty" solution.
nftw is just not a well designed C interface.
Related
The move to the new continuations API in Arrow brought with it a handy new function: shift, in theory letting me get rid of ensure(false) { NewError() } or NewError().left().bind() constructs.
But I'm not sure how to properly use it. The documentation states that it is intended to short-circuit the continuation, and there are no conditionals, so it should always take the parameter, and (in either parlance) "make it a left value", and exit the scope.
So what is the type parameter B intended to be used for? It determines the return type of shift, but shift will not return. Given no more context, B can not be inferred, leading to this kind of code:
val res = either {
val intermediate = mayReturnNull()
if (intermediate == null) {
shift<Nothing>(IntermediateWasNull())
}
process(intermediate)
}
Note the <Nothing> (and ignore the contrived example, the main point is that shifts return type can not be inferred – the actual type parameter does not even matter).
I could wrap shift like this:
suspend fun <L> EffectScope<L>.fail(left: L): Nothing = shift(left)
But I feel like that is missing the point. Any explanations/hints would be greatly appreciated.
That is a great question!
This is more a matter of style, ideally we'd have both but they conflict so we cannot have both APIs available.
So shift always returns Nothing in its implementation, and so the B parameter is completely artificial.
This is something that is true for a lot of other things in Kotlin, such as object EmptyList : List<Nothing>. The Kotlin Std however exposes it as fun <A> emptyList(): List<A> = EmptyList.
For Arrow to stay consistent with APIs found in Kotlin Std, and to remain as Kotlin idiomatic as possible we also require a type argument just like emptyList. This has been up for discussion multiple times, and the Kotlin languages authors have stated that it was decided too explicitly require A for emptyList since that results in the best and most consistent ergonomics in Kotlin.
In the example you shared I would however recommend using ensureNotNull which will also smart-cast intermediate to non-null.
Arrow attempts to build the DSL so that you don't need to rely on shift in most cases, and you should prefer ensure and ensureNotNull when possible.
val res = either {
val intermediate = mayReturnNull()
ensureNotNull(intermediate) { IntermediateWasNull() }
process(intermediate) // <-- smart casted to non-null
}
I have a logging function in a Swift class of mine with the following signature:
fun logDetails(msg: String, function: String = #function, file: String = #file, line: Int = #line) {
I like the fact that I can call it from anywhere in my code, and it will fill in the function, file and source code line where it was called.
Is there any way to accomplish this in Kotlin?
There is no direct language feature in Kotlin (that I know of) that could do what you want in a similar way.
The only thing along those lines that I can think of in Kotlin/JVM (or Java) is to use Thread.currentThread().getStackTrace(). This returns an array of StackTraceElements that represent the call stack of the current thread. Each of those elements can provide information about filename, class name, method name and line number.
The problem is that computing such stack traces is costly, so I wouldn't use that extensively for extensive logging purposes (unless you're trying to temporarily debug something without access to a debugger).
I know there was documented in main kotlin page, but there is no clear explanation about when to use it, why this function need a receiver as a function. What would be the correct way to create a correct definition of inline function.
This is inline function
inline fun String?.toDateString(rawDateFormat: String = MMMM_DD_YYYY, outputDate: String = MM_DD_YYYY, block: (date: String) -> String): String {
return try {
var sdf = SimpleDateFormat(rawDateFormat, Locale.US)
val date = sdf.parse(this.orEmpty())
sdf = SimpleDateFormat(outputDate, Locale.US)
block(sdf.format(date ?: Date()).orEmpty())
} catch (ex: Exception) {
block("")
}
}
The same way we also can do
inline fun String?.toDateString(rawDateFormat: String = MMMM_DD_YYYY, outputDate: String = MM_DD_YYYY): String {
return try {
var sdf = SimpleDateFormat(rawDateFormat, Locale.US)
val date = sdf.parse(this.orEmpty())
sdf = SimpleDateFormat(outputDate, Locale.US)
sdf.format(date ?: Date()).orEmpty()
} catch (ex: Exception) {
""
}
}
If anyone could have a detail explanation about this?
Edit:
I understand that the inline function will insert the code whenever it called by the compiler. But this come to my attention, when I want to use inline function without functional parameter receiver type the warning show as this in which should have a better explain. I also want to understand why this is such recommendation.
There are few things here.
First, you ask about using a function with a receiver. In both cases here, the receiver is the String? part of String?.toDateString(). It means that you can call the function as if it were a method of String, e.g. "2021-01-15 12:00:00".toDateString(…).
The original String? is accessible as this within the function; you can see it in the sdf.parse(this.orEmpty()) call. (It's not always as obvious as this; you could simply call sdf.parse(orEmpty()), where the this. is implied.)
Then you ask about inline functions. All you have to do is to mark the function as inline, and the compiler will automatically insert its code wherever it's called, instead of defining a function in the usual way. But you don't need to worry about how it's implemented; there are just a few visible effects in the code. In particular, if a function is inline and accepts a function parameter, then its lambda can do a few things (such as calling return) that it couldn't otherwise do.
Which leads us to what I think is your real question: about the block function parameter. Your first example has this parameter, with the type (date: String) -> String — i.e. a function taking a single String parameter and returning another String. (The technical term for this is that toDateString() is a higher-order function.)
The toDateString() function calls this block function before returning, applying it to the date string it has formatted before returning it to the caller.
As to why it does this, it's hard to tell. That's why we put documentation comments before functions: to explain anything that's not obvious from the code! Ideally, there would be a comment explaining why you're required to supply a block lamdba (or function reference), when it's not vital to what the function does.
There are times when blocks passed this way are very useful. For example, the joinToString() function accepts an optional transform parameter, which it applies to each item before joining it to the list. If it didn't, the effect would be a lot more awkward to obtain. (You'd probably have to apply a map() to the collection before calling joinToString(), which would be less efficient.)
But this isn't one of those times. As your second example shows, toDateString() would work perfectly well without the block parameter — and then if you needed to pass the result through another function, you could just call it on toDateString()'s result.
Perhaps if you included a link to the ‘main kotlin page’ where you saw this, it might give some more context?
The edited question also asks about the IDE warning. This is shown when it thinks inlining a function won't give a significant improvement.
When no lambdas are involved, the only potential benefit from inlining a function is performance, and that's a trade-off. It might avoid the overhead of a function call wherever it's called — but the Java runtime will often inline small functions anyway, all on its own. And having the compiler do the inlining comes at the cost of duplicating the function's code everywhere it's called; the increased code size is less likely to fit into memory caches, and less likely to be optimised by the Java runtime — so that can end up reducing the performance overall. Because this isn't always obvious, the IDE gives a warning.
It's different when lambdas are involved, though. In that case, inlining affects functionality: for example, it allows non-local returns and reified type parameters. So in that case there are good reasons for using inline regardless of any performance implications, and the IDE doesn't give the warning.
(In fact, if a function calls a lambda it's passed, inlining can have a more significant performance benefit: not only does the function itself get inlined, but the lambda itself usually does as well, removing two levels of function call — and the lambda is often called repeatedly, so there can be a real saving.)
Example:
data class T(val flag: Boolean) {
constructor(n: Int) : this(run {
// Some computation here...
<Boolean result>
})
}
In this example, the custom constructor needs to run some computation in order to determine which value to pass to the primary constructor, but the compiler does not accept the run, citing Cannot access 'run' before superclass constructor has been called, which, if I understand correctly, means instead of interpreting it as the non-extension run (the variant with no object reference in https://kotlinlang.org/docs/reference/scope-functions.html#function-selection), it construes it as a call to this.run (the variant with an object reference in the above table) - which is invalid as the object has not completely instantiated yet.
What can I do in order to let the compiler know I mean the run function which is not an extension method and doesn't take a scope?
Clarification: I am interested in an answer to the question as asked, not in a workaround.
I can think of several workarounds - ways to rewrite this code in a way that works as intended without calling run: extracting the code to a function; rewriting it as a (possibly highly nested) let expression; removing the run and invoking the lambda (with () after it) instead (funnily enough, IntelliJ IDEA tags that as Redundant lambda creation and suggests to Inline the body, which reinstates the non-compiling run). But the question is not how to rewrite this without using run - it's how to make run work in this context.
A good answer should do one of the following things:
Explain how to instruct the compiler to call a function rather than an extension method when a name is overloaded, in general; or
Explain how to do that specifically for run; or
Explain that (and ideally also why) it is not possible to do (ideally with supporting references); or
Explain what I got wrong, in case I got something wrong and the whole question is irrelevant (e.g. if my analysis is incorrect, and the problem is something other than the compiler construing the call to run as this.run).
If someone has a neat workaround not mentioned above they're welcome to post it in a comment - not as an answer.
In case it matters: I'm using multi-platform Kotlin 1.4.20.
Kotlin favors the receiver overload if it is in scope. The solution is to use the fully qualified name of the non-receiver function:
kotlin.run { //...
The specification is explained here.
Another option when the overloads are not in the same package is to use import renaming, but that won't work in this case since both run functions are in the same package.
I was having a look at the Arrow library found here. Why would ever want to use an Option type instead of Kotlin's built in nullables?
I have been using the Option data type provided by Arrow for over a year, and there at the beginning, we did the exact same question to ourselves. The answer follows.
Option vs Nullable
If you compare just the option data type with nullables in Kotlin, they are almost even. Same semantics (there is some value or not), almost same syntax (with Option you use map, with nullables you use safe call operator).
But when using Options, you enable the possibility to take benefits from the arrow ecosystem!
Arrow ecosystem (functional ecosystem)
When using Options, you are using the Monad Pattern. When using the monad pattern with libraries like arrow, scala cats, scalaz, you can take benefits from several functional concepts. Just 3 examples of benefits (there is a lot more than that):
1. Access to other Monads
Option is not the only one! For instance, Either is a lot useful to express and avoid to throw Exceptions. Try, Validated and IO are examples of other common monads that help us to do (in a better way) things we do on typical projects.
2. Conversion between monads + abstractions
You can easily convert one monad to another. You have a Try but want to return (and express) an Either? Just convert to it. You have an Either but doesn't care about the error? Just convert to Option.
val foo = Try { 2 / 0 }
val bar = foo.toEither()
val baz = bar.toOption()
This abstraction also helps you to create functions that doesn't care about the container (monad) itself, just about the content. For example, you can create an extension method Sum(anyContainerWithBigDecimalInside, anotherContainerWithBigDecimal) that works with ANY MONAD (to be more precise: "to any instance of applicative") this way:
fun <F> Applicative<F>.sum(vararg kinds: Kind<F, BigDecimal>): Kind<F, BigDecimal> {
return kinds.reduce { kindA, kindB ->
map(kindA, kindB) { (a, b) -> a.add(b) }
}
}
A little complex to understand, but very helpful and easy to use.
3. Monad comprehensions
Going from nullables to monads is not just about changing safe call operators to map calls. Take a look at the "binding" feature that arrow provides as the implementation of the pattern "Monad Comprehensions":
fun calculateRocketBoost(rocketStatus: RocketStatus): Option<Double> {
return binding {
val (gravity) = rocketStatus.gravity
val (currentSpeed) = rocketStatus.currentSpeed
val (fuel) = rocketStatus.fuel
val (science) = calculateRocketScienceStuff(rocketStatus)
val fuelConsumptionRate = Math.pow(gravity, fuel)
val universeStuff = Math.log(fuelConsumptionRate * science)
universeStuff * currentSpeed
}
}
All the functions used and also the properties from rocketStatus parameter in the above example are Options. Inside the binding block, the flatMap call is abstracted for us. The code is a lot easier to read (and write) and you don't need to check if the values are present, if some of them is not, the computation will stop and the result will be an Option with None!
Now try to imagine this code with null verifications instead. Not just safe call operators but also probably if null then return code paths. A lot harder isn't it?
Also, the above example uses Option but the true power about monad comprehensions as an abstraction is when you use it with monads like IO in which you can abstract asynchronous code execution in the exact same "clean, sequential and imperative" way as above :O
Conclusion
I strongly recommend you to start using monads like Option, Either, etc as soon as you see the concept fits the semantics you need, even if you are not sure if you will take the other big benefits from the functional ecosystem or if you don't know them very well yet. Soon you'll be using it without noticing the learning-curve. In my company, we use it in almost all Kotlin projects, even in the object-oriented ones (which are the majority).
Disclaimer: If you really want to have a detailed talk about why Arrow is useful, then please head over to https://soundcloud.com/user-38099918/arrow-functional-library and listen to one of the people who work on it. (5:35min)
The people who create and use that library simple want to use Kotlin differently than the people who created it and use "the Option datatype similar to how Scala, Haskell and other FP languages handle optional values".
This is just another way of defining return types of values that you do not know the output of.
Let me show you three versions:
nullability in Kotlin
val someString: String? = if (condition) "String" else null
object with another value
val someString: String = if (condition) "String" else ""
the Arrow version
val someString: Option<String> = if (condition) Some("String") else None
A major part of Kotlin logic can be to never use nullable types like String?, but you will need to use it when interopting with Java. When doing that you need to use safe calls like string?.split("a") or the not-null assertion string!!.split("a").
I think it is perfectly valid to use safe calls when using Java libraries, but the Arrow guys seem to think different and want to use their logic all the time.
The benefit of using the Arrow logic is "empowering users to define pure FP apps and libraries built atop higher order abstractions. Use the below list to learn more about Λrrow's main features".
One thing other answers haven't mentioned: you can have Option<Option<SomeType>> where you can't have SomeType??. Or Option<SomeType?>, for that matter. This is quite useful for compositionality. E.g. consider Kotlin's Map.get:
abstract operator fun get(key: K): V?
Returns the value corresponding to the given key, or null if such a key is not present in the map.
But what if V is a nullable type? Then when get returns null it can be because the map stored a null value for the given key or because there was no value; you can't tell! If it returned Option<V>, there wouldn't be a problem.