C#•3y ago

Need help understanding lambda statements and delegates

I'm extremely confused about the syntax and it's hemorraging my ability to work on other stuff (LINQ, middleware) How do I even begin comprehending the syntax? I keep rereading the MSDN documentation and watching youtube videos and nothing makes sense

108 Replies

ero•3y ago

What exactly is confusing you? You can write any lambda as a full method, if that helps?

Func<string, int> parse = input => int.Parse(input);

Func<string, int> parse = input => int.Parse(input);

int Parse(string input)
{
  return int.Parse(input);
}

int Parse(string input)
{
  return int.Parse(input);
}

canton7•3y ago

More details please. What exactly is confusing you?

surwrenOP•3y ago

Yea so why does this look utterly different from the middleware lambda?

app.Use(async (context, next) =>...

app.Use(async (context, next) =>...

I know a lambda basically lacks a method name It's just parameters => dosomething

ero•3y ago

async Task<PotentialReturnType> Lambda(ContextType context, NextType next)
{
  // lambda body
}

async Task<PotentialReturnType> Lambda(ContextType context, NextType next)
{
  // lambda body
}

surwrenOP•3y ago

But why is the syntax so drastically different between lambdas used in middleware, delegate, and other functions?

ero•3y ago

different in what way?

surwrenOP•3y ago

Func<string, int> parse = input =>

Func<string, int> parse = input =>

app.use (async (context,next) =>

app.use (async (context,next) =>

The parameters aren't even lined up the same way I don't get it In the first one the return type is int and input is string But they are both inside of func <> And then in the second, both context and next are input parameters and there is no return type specified in the lambda Why is there no consistency between the syntaxes? There are no patterns Or at least my stupid brain can't see the patterns I can only recognize that => indicates to dosomething But i keep getting tripped up by the lambda signature It seems that the signature changes every time it's used and it drives me crazy because I can't figure out why Ok for example

parse=input

parse=input

what does that even mean?

ero•3y ago

parse is the lambda's name just like any other variable int myInt = 0; string myString = ""; i could have named it myLambda if you prefer that input is the string parameter's name just like in the full method body below string input i could have wrapped input in parens if you prefer that Func<string, int> myLambda = (input) => int.Parse(input); what you're using in app.Use() is an anonymous function, one which doesn't have a name

surwrenOP•3y ago

But why is input even there, and missing a type declaration?

ero•3y ago

so all that remains is (input) => int.Parse(input); the type declaration is in the Func type Func<string, int> takes in one string parameter, and returns one integer

surwrenOP•3y ago

So the return type of a lambda is always the last value of the

Func

Func

declaration?

ero•3y ago

indeed

surwrenOP•3y ago

Func<int, int, bool> testForEquality = (x, y) => x == y;

Func<int, int, bool> testForEquality = (x, y) => x == y;

Return would be bool

Func<int,int[],List<int>,List<int>>

Func<int,int[],List<int>,List<int>>

Return would be List<int>?

ero•3y ago

mhm

surwrenOP•3y ago

So then the variable names (x, y) correspond in order with the int,int declarations in Func (for the first lambda)?

ero•3y ago

yup

surwrenOP•3y ago

If it was Func<int,string,int> trial = (a,b)=> It would be int a, string b

ero•3y ago

yup

surwrenOP•3y ago

I still dont get this

ero•3y ago

you can actually declare them explicitly

Func<int, string, int> foo = (int a, string b) => { };

Func<int, string, int> foo = (int a, string b) => { };

surwrenOP•3y ago

Ah But it needs to match the Func<> declaration?

ero•3y ago

yeah

surwrenOP•3y ago

Im still having trouble mapping the app.use one in my head

ero•3y ago

well, just like you don't need to do

string foo = "Hello, world!";
Console.WriteLine(foo);

string foo = "Hello, world!";
Console.WriteLine(foo);

but instead can just skip the variable declaration;

Console.WriteLine("Hello, world!");

Console.WriteLine("Hello, world!");

the same goes for lambdas

Func<int, string, int> foo = (bar, baz) => int.Parse(baz) - bar;

SomeMethodThatTakesALambdaOfTheAboveSignature(foo);

Func<int, string, int> foo = (bar, baz) => int.Parse(baz) - bar;

SomeMethodThatTakesALambdaOfTheAboveSignature(foo);

SomeMethodThatTakesALambdaOfTheAboveSignature((bar, baz) => int.Parse(baz) - bar);

SomeMethodThatTakesALambdaOfTheAboveSignature((bar, baz) => int.Parse(baz) - bar);

surwrenOP•3y ago

Ohhh so you excluded the

foo

foo

because inserting the lambda directly into the function doesn't require you to give the lambda a variable name But async keyword remains?

ero•3y ago

the async is there because the return type of your lambda is a Task or a Task<T> (depending on if it returns anything) and Tasks can be awaited

surwrenOP•3y ago

I have no idea what a Task<T> is

ero•3y ago

you're gonna have to learn that if you're gonna wanna do web dev

surwrenOP•3y ago

Oh msdn documentation shows you can do Task<int>

ero•3y ago

T would be int in that case, yes

surwrenOP•3y ago

I feel like I need a whole different thread for clearing up understanding on Tasks Ik Task is a return type in itself But thats sort of digressimg Ok back to lambdas

app.Use(async (context, next) =>
{
    if (context.Request.Path.StartsWithSegments("/Login/"))
    {
        await next(context);
        return;
    }

    string sessionId = context.Request.Cookies["sessionId"];
    if (sessionId == null)
    {
        
        context.Response.Redirect("/Login/");
    }
    else
    {
        await next(context);
    }
});

app.Use(async (context, next) =>
{
    if (context.Request.Path.StartsWithSegments("/Login/"))
    {
        await next(context);
        return;
    }

    string sessionId = context.Request.Cookies["sessionId"];
    if (sessionId == null)
    {
        
        context.Response.Redirect("/Login/");
    }
    else
    {
        await next(context);
    }
});

In this case the entire {} chunk is the dosomething part right

ero•3y ago

yeah

surwrenOP•3y ago

And the entire lambda is just thrown into app.use Hmm Lemme grab the delegates code i was confused about for 2 months Gimme a bit ok //defining a new datatype called IntOps that represent method that accept int and return an int

delegate int IntOps(int n);

delegate int IntOps(int n);

//Once we declare this, we can create a method that uses this delegate type

void ApplyOperation(int[] arr, IntOps ops)

void ApplyOperation(int[] arr, IntOps ops)

//we can use the parameter just like a normal method in our method

static void ApplyOperation(int[] arr, IntOps ops){
for (int i=0; i<arr.Length; i++) {
arr[i] = ops(arr[i]);
}
}

static void ApplyOperation(int[] arr, IntOps ops){
for (int i=0; i<arr.Length; i++) {
arr[i] = ops(arr[i]);
}
}

//method Add10 match the signature of the IntOps delegate //I don't understand what this means, or why

static void Main() {
int[] A = new int[] {1, 2, 3};
IntOps myOp = Add10;
ApplyOperation(A, myOp);
//A should contain {11, 12, 13} now.
}
static int Add10(int x) {
return x + 10;
}

static void Main() {
int[] A = new int[] {1, 2, 3};
IntOps myOp = Add10;
ApplyOperation(A, myOp);
//A should contain {11, 12, 13} now.
}
static int Add10(int x) {
return x + 10;
}

//Calling the method that accepts a delegate (using anonymous method)

static void Main() {
int[] A = new int[] {1, 2, 3};
IntOps myOp = delegate(int a)
{
return a + 10;
};
ApplyOperation(A, myOp);
//A should contain {11, 12, 13} now.
}

static void Main() {
int[] A = new int[] {1, 2, 3};
IntOps myOp = delegate(int a)
{
return a + 10;
};
ApplyOperation(A, myOp);
//A should contain {11, 12, 13} now.
}

//Calling the method that accepts a delegate (using delegate)

static void Main() {
int[] A = new int[] {1, 2, 3};
IntOps myOp = (int a) => { return a + 10; }; 
ApplyOperation(A, myOp);
//A should contain {11, 12, 13} now.
}

static void Main() {
int[] A = new int[] {1, 2, 3};
IntOps myOp = (int a) => { return a + 10; }; 
ApplyOperation(A, myOp);
//A should contain {11, 12, 13} now.
}

Ig here my issue is, what is an anonymous function and what does it have to do with lambdas? Ok so if we compare the one I just posted to yours (with explicit declaration)

IntOps myOp = (int a) => { return a + 10; };

IntOps myOp = (int a) => { return a + 10; };

Func<string, int> parse = (string input) => { int.Parse(input) };

Func<string, int> parse = (string input) => { int.Parse(input) };

Oh so it's replacing the Func<> declaration with IntOps myOp But why? What even is IntOps myOp and why does it apply a lambda

surwrenOP•3y ago

Ok, so

delegate int IntOps(int n);

delegate int IntOps(int n);

Is declaring a delegate that returns int, of name IntOps, that accepts parameter int n https://www.tutorialsteacher.com/csharp/csharp-delegates Based on the graphic in this thread

C# Delegates

surwrenOP•3y ago

--- Oh so it's just showing 3 different ways of assigning a method to

IntOps myOp

IntOps myOp

First way is by assigning the method name directly (Add10) Second way is by assigning an anonymous method (I don't really get this) Third way is assigning using a lambda === Ok so in some videos they refer to lambda functions as anonymous functions. But it seems that in this code example they are not the same? Anonymous function example

IntOps myOp = delegate(int a) 
{
return a+10
}

IntOps myOp = delegate(int a) 
{
return a+10
}

Lambda expression example

IntOps myOp = (int a) => { return a+10; };

IntOps myOp = (int a) => { return a+10; };

@Ero are anonymous functions the same thing as lambdas? If they are, why does the syntax differ?

canton7•3y ago

There is old and new syntax: old: delegate (int x) { return ...;}, new: x => ... The => syntax is called lambda syntax

surwrenOP•3y ago

So what is the delegate(int x) syntax called

canton7•3y ago

Have you played with sharplab?

surwrenOP•3y ago

Nope

canton7•3y ago

It's probably worth playing with. https://sharplab.io

SharpLab

C#/VB/F# compiler playground.

canton7•3y ago

See what the different syntaxes generate They all end up generating a normal C# method under the hood

surwrenOP•3y ago

That would be great for actually learning yeah but I'm trying to understand what the delegate(int x) syntax is called Does it have a name?

canton7•3y ago

I don't know if it had a specific name. You used the syntax to declare what was called an anonymous function mainly, iirc. There are lots of names for the same things

surwrenOP•3y ago

in my notes it's labelled an anonymous function in the videos i've seen they refer to lambdas as anonymous functions. Are anonymous functions == lambdas or are anonymous functions an umbrella term that includes lambdas?

canton7•3y ago

Iambdas, anonymous methods, closures, all effectively the same thing

surwrenOP•3y ago

If they are the same thing then why does the syntax not have => Why does it have delegate() instead Does this mean delegate() is a method used to designate a lambda/anon method/closure?

canton7•3y ago

strictly speaking in C#, the => syntax is a lambda. You can use that to declare an anonymous function/closure, but you can also use => to declare a linq expression (though that's an added complication you can probably ignore for now) The old and new syntaxes do the same thing

surwrenOP•3y ago

Oh yeah I did see linQ expressions with ()=> I wanted to ask about those too I was really confused when I first saw them

canton7•3y ago

The complication there linq is in two parts: you can use the same => syntax to declare both an inline method, and something which an be translated to SQL. But let's ignore that for now

surwrenOP•3y ago

How do I differentiate between the two?

canton7•3y ago

All these do the same thing:

public int Method(int x) { return x * 2; }
Func<int, int> doubler = new Func<int, int>(Method);
Func<int, int> doubler = Method;
Func<int, int> doubler = delegate(int x) { return x*2; };
Func<int, int> doubler = x => x * 2;

public int Method(int x) { return x * 2; }
Func<int, int> doubler = new Func<int, int>(Method);
Func<int, int> doubler = Method;
Func<int, int> doubler = delegate(int x) { return x*2; };
Func<int, int> doubler = x => x * 2;

In the first two, we're writing a method ourselves and creating the delegate instance doubler from that method. In the latter two, we're using different syntax to write the method inline, and the compiler is creating the method for us under the hood, and then creating the delegate instance from that method You don't see the 3rd syntax any more -- the 4th one is a replacement which is better

surwrenOP•3y ago

Whats the purpose of doing = new Func<int,int>(Method) if = Method achieves the same thing?

canton7•3y ago

The way of writing the method inline is variously called a closure, lambda, inline method, anonymous delegate... All basically the same thing, coming from different languages and paradigmes = new DelegateType(Method) came first, then = Method was added later as a shorthand to reduce the boilerplate. It's still very occasionally useful = Method still generates new DelegateType(Method) under the hood

surwrenOP•3y ago

Func<> is new DelegateType? Why does c# name the methods relating to lambdas things that begin with 'delegate..'? I thought delegates and lambdas were separate concepts

canton7•3y ago

Yes. The framework declares it as public delegate TResult Func<in T, out TResult>(T arg);: https://source.dot.net/#System.Private.CoreLib/src/libraries/System.Private.CoreLib/src/System/Function.cs,7 A delegate is its own thing, similar to but separate from a class and struct. A delegate is a thing which points to a method, and can be used to invoke the method. You declare the delegate type with delegate returntype DelegateType(param1, param2, ...) (and the type describes the delegate's parameters and return type), and instantiate it with new DelegateType(method) (although as we've seen, the compiler will add the instantiation for you to reduce boilerplate most of the time)

surwrenOP•3y ago

Oh so lambdas are just one way of assigning a method to a delegate/method pointer?

canton7•3y ago

They're a way of declaring a method inline, and creating a new delegate instance which points to that method, yes

surwrenOP•3y ago

Are delegates ref type or value type? Pointer seems to indicate ref type

canton7•3y ago

They're immutable reference types Confusingly, they can be combined: SomeDelegate result = delegate1 + delegate2. Invoking result will invoke both delegate1 and delegate2. Events are built around that bit of functionality, although you don't see it used outside of events

surwrenOP•3y ago

I haven't covered events yet, probably need to cover linQ when I wake up in the morning Will be back with more questions about lambda syntaxes but in linq

canton7•3y ago

https://sharplab.io/#v2:EYLgHgbALANAJiA1AHwAICYCMBYAUKgZgAIMiBhIgbzyNpONSiIFkAKASiprp9QFYAPAEsAdgBcYRUWIB8ROAHsArsAA2AUwBORALxEwuuQYBURdAG5utAL55rQA. See there. There's some extra stuff around caching the delegate instance, but you should be able to see what's goind on

SharpLab

C#/VB/F# compiler playground.

surwrenOP•3y ago

Wdym by caching 😳

canton7•3y ago

It put the method x * 2 (which it called <M>b__0_0) on a new generated inner class called <>c, and then did new Func<int, int>(<>c.<>9.<M>b__0_0) So if the code Func<int, int> doubler = x => x * 2; is run lots of times, it only needs to create one delegate instance, which makes it cheaper https://en.wikipedia.org/wiki/Cache_(computing)

surwrenOP•3y ago

Oh you mean the compiler adding the delegate instantiation boilerplate and then caching the resulting delegate Bruh

canton7•3y ago

yep

surwrenOP•3y ago

Guh I wish I could add you so it would be easier to reach than pinging you in this thread 🥲 Ah, so this

IntOps myOp = (int a) => { return a + 10; };

IntOps myOp = (int a) => { return a + 10; };

requires IntOps to define the return and input param type in

delegate int IntOps(int n)

delegate int IntOps(int n)

because there's no Func<int,int> definition

static void ApplyOperation(int[] arr, IntOps ops){
for (int i=0; i<arr.Length; i++) {
arr[i] = ops(arr[i]);
}
}

static void ApplyOperation(int[] arr, IntOps ops){
for (int i=0; i<arr.Length; i++) {
arr[i] = ops(arr[i]);
}
}

And then this code just accepts the lambda assigned to IntOps as an input parameter, which it then manipulates int[] arr with === Then I have a question, what is the difference between these two?

Func<int,int> SomeMethod

Func<int,int> SomeMethod

and

delegate int SomeMethod (int n)

delegate int SomeMethod (int n)

you can assign a corresponding lambda statement to both of these. What purpose does each serve?

Kouhai•3y ago

There is no real difference These are is the definitions of Func from the official repo https://github.com/dotnet/runtime/blob/57bfe474518ab5b7cfe6bf7424a79ce3af9d6657/src/libraries/System.Private.CoreLib/src/System/Function.cs

GitHub

runtime/Function.cs at 57bfe474518ab5b7cfe6bf7424a79ce3af9d6657 · d...

.NET is a cross-platform runtime for cloud, mobile, desktop, and IoT apps. - runtime/Function.cs at 57bfe474518ab5b7cfe6bf7424a79ce3af9d6657 · dotnet/runtime

surwrenOP•3y ago

Both work when inserted into

void ApplyOperation(int[] arr, Func<int, int> ops) ...

void ApplyOperation(int[] arr, Func<int, int> ops) ...

void ApplyOperation(int[] arr, IntOps ops) ...

void ApplyOperation(int[] arr, IntOps ops) ...

Just that to use a delegate function, you need another line

delegate int IntOps(int n);

delegate int IntOps(int n);

lol it supports up to a max of 16 input params

Kouhai•3y ago

Yeah 😅

surwrenOP•3y ago

well the Func<> syntax certainly seems more convenient considering that you don't need to declare a

delegate returntype methodname(input params)

delegate returntype methodname(input params)

but is there any advantage to using the delegate int IntOps(int n) method at all?

Kouhai•3y ago

I don't know about others, but I personally use Func<> or Action<> most of the times unless I want to have a type name.

surwrenOP•3y ago

wdym by type name

Kouhai•3y ago

Very rarely you might use it for ref, out because Func and Action don't allow that

surwrenOP•3y ago

you mean like

delegate int IntOps(ref int n)

delegate int IntOps(ref int n)

Kouhai•3y ago

Basically instead of public void Test(Func<int, int, int, string> callback) public delegate string CallbackDelegate(int param0, int param1, int param2) and public void Test(CallbackDelegate callback) Yeah, with Func and Action you can't ref and out

surwrenOP•3y ago

gdi are you talking about delegate name? I still don't quite understand

Kouhai•3y ago

Oh sorry, I didn't clarify, yes I was taking about the the delegate name It might be easier for the caller to understand the purpose of it. Or it might save you typing extra characters if you use that delegate a lot in your code

surwrenOP•3y ago

I also have a question- when using top level statements (TLS), it seems that the delegate variable holding the relevant lambda can be declared at any point in the code (before or after the method calling it) Why is this possible? Whereas other variable declarations must occur in the proper scope ahead of the TLS call

Kouhai•3y ago

Wdym, I don't think you can call it before the variable is declared 🤔

surwrenOP•3y ago

I'll screengrab when I get back to the computer How would I go about looking for the definitions of app.Use from the original asp repo?

Kouhai•3y ago

app?

surwrenOP•3y ago

Middlewares

Kouhai•3y ago

Oh Here's the default implementation for IApplicationBuilder

Kouhai•3y ago

https://github.com/dotnet/aspnetcore/blob/51fc6824e30016404964fc96a4a098a764bdbbfa/src/Http/Http/src/Builder/ApplicationBuilder.cs

GitHub

aspnetcore/ApplicationBuilder.cs at 51fc6824e30016404964fc96a4a098a...

ASP.NET Core is a cross-platform .NET framework for building modern cloud-based web applications on Windows, Mac, or Linux. - aspnetcore/ApplicationBuilder.cs at 51fc6824e30016404964fc96a4a098a764b...

surwrenOP•3y ago

it's callable

Kouhai•3y ago

It's declared before getting called, no?

surwrenOP•3y ago

wait if that's the declaration then what does this do

delegate int Testing(int n);

delegate int Testing(int n);

Kouhai•3y ago

Oh, that's the type declaration, this is similar to class MyClass {}

surwrenOP•3y ago

ohhh is it correct to call it the class declaration? the syntax is green indicating that Testing it's a class/object

Kouhai•3y ago

well, it's a delegate declaration not a class

surwrenOP•3y ago

what does green coloration in the IDE indicate object?

Kouhai•3y ago

I think with the default visual studio config it indicates a Type

surwrenOP•3y ago

so in other word, classes and delegates are types?

surwrenOP•3y ago

Kouhai•3y ago

A class is a type, let's think of it that way when you do class MyClass {} And instantiate it like this var myClassInstance = new MyClass() what's the type of myClassInstance?

surwrenOP•3y ago

but if you assign a variable to a type, it's not green

Kouhai•3y ago

Try Int32 iUse = 10 What is it's color?

surwrenOP•3y ago

Green, but it's a different shade from the object/delegate green

surwrenOP•3y ago

Int32 is a struct

Kouhai•3y ago

Yup, it's a struct, and the reason it has a darker shade is visual studio is suggesting to substitute it with int int is a keyword in c#, that keyword map directly to the Type Int32

surwrenOP•3y ago

oh so blue indicates keywords

Kouhai•3y ago

That's right 👍

surwrenOP•3y ago

that's so fascinating wtf why are read only types like int considered keywords?

Kouhai•3y ago

because they are defined as keywords in ECMA-334 (the standard specs for C#) And it's more intuitive to just do int short instead of Int32 Int16

Gaming

Programming

Need help understanding lambda statements and delegates

Did you find this page helpful?