Default implementations

[Pyrdacor]

The idea is to allow a short language syntax to auto-implement default behaviors by the compiler. This proposal mainly focuses on functions. What comes to mind are things like GetHashCode or Clone. For me the latter is quite useful as currently there is no good way to clone objects (especially deep cloning).

With records we now have classes with some default behavior auto-implemented already. So something similar was done before.

However the new keyword allows more flexible approaches and more control. And it could in theory be used for any type of object.

class MyClass : ICloneable
{
    public int MyProp { get; set; }

    default object Clone();
    default int GetHashCode();
}

In this case the compiler would generate something like this:

class MyClass : ICloneable
{
    public int MyProp { get; set; }

    public object Clone()
    {
        return new MyClass
        {
            MyProp = MyProp
        };
    }

    public int GetHashCode()
    {
        return HashCode.Combine(MyProp);
    }
}

It might also work with patterns like IDisposable or INotifyPropertyChanged. So basically for well-known patterns which have a default behavior most of the time. This saves a lot of time and boilerplate code. Especially cloning and hash code generation would be enough to justify such language feature imo.

So the accepted functions might be limited by the compiler and it can generate an error if the function has no default behavior. This can be extended later if needed.

This idea can of course be extended or adjusted. My goal is to allow cloning easier and save boilerplate code for classes.

[HaloFour]

Feels like something better suited to a source generator, one that can be configurable and apply policies to the emitted implementation of those methods.

[iam3yal]

@Pyrdacor

this is no real solution.

With all due respect we see this a lot here, the moment someone mentions SG or 3rd party packages some guys here go but this is a "workaround", "not a real solution" or whatever but if you REALLY need it then YOU should definitely create it and maintain it yourself, don't you think? because the mechanism exists but some of you just don't want to do it and expect other devs to do it for you.

So I'll say this if it's not a real solution you probably don't have a real problem.

[Pyrdacor]

I didn't say that I couldn't implement it myself. But if the implementation is only inside my own 3rd party nuget package, this is not really useful to have a general deep cloning mechanism in C#.

[Pyrdacor]

So I guess nobody.

That means it will be you. :)

Wow didn't expect insults in a place like this, but ok.

[TahirAhmadov]

I didn't say that I couldn't implement it myself. But if the implementation is only inside my own 3rd party nuget package, this is not really useful to have a general deep cloning mechanism in C#.

Deep cloning is not really a well defined set of requirements. I struggle with it in my code, and that's just one code base (admittedly, with many many solutions). In some cases, there are recursive references, in others, some classes need to be excluded from being cloned due to business requirements, etc. It's definitely not something which can be added to the language for everybody's usage.

Wow didn't expect insults in a place like this, but ok.

You misunderstood my comment. I didn't mean to say "you are a nobody", what I meant was "since nobody else is willing to do it, then it will have to be you". Basically the same thing that @iam3yal said:

if you REALLY need it then YOU should definitely create it and maintain it yourself, don't you think? because the mechanism exists but some of you just don't want to do it and expect other devs to do it for you.

Believe me, neither me nor anybody else here are trying to insult you.

[Pyrdacor]

@Pyrdacor

this is no real solution.

With all due respect we see this a lot here, the moment someone mentions SG or 3rd party packages some guys here go but this is a "workaround", "not a real solution" or whatever but if you REALLY need it then YOU should definitely create it and maintain it yourself, don't you think? because the mechanism exists but some of you just don't want to do it and expect other devs to do it for you.

So I'll say this if it's not a real solution you probably don't have a real problem.

Maybe a few words about that. It's not about laziness in those cases. It's about trust in the expertise. I could also repair my car myself and I do it regularly if it's a small thing. But for most things I go to my nearby car repair. I could also ask a friend who says he can repair cars, but in general I still go to the experts for that.

3rd party libraries often bring this feeling with them, that the author didn't do it the best way. Having features inside the runtime or compiler feels safe and everybody thinks that it is most likely implemented in the best possible way. Moreover it is much more reviewed and discussed. And there are many people behind it so that maintenance is guaranteed.

In the end everybody in these discussions wants to improve something. I don't think that most people just are lazy and want others to do their work.