Article c++

`defaulted` constructor in C++

Lu Pan

05 Dec 2021 • 1 min read

You have seen Foo() = default;. It declares a default constructor (duh!). But what does it really do, and when is it actually useful? "When is it useful" is a very interesting question in my opinion because next time when you see = default; in other people's code, you would understand why it's there. It's all about communicating intentions and making code self documented.

First of all, a default constructor is the one that gets called when no arguments are passed in – T{}. For class/struct T, if you have constructor(s) of any kind defined explicitly, compiler will stay out of your way; you get what you see.

class Foo {
    // as long as there is at least one user-defined ctor
    // compiler will not do anything about ctors (including
    // default ctor)
    Foo(int a) {}
};

If no constructor is defined by user, compiler will declare one for you. The invariant is that any class have at least one constructor declaration.

class Foo {
  // compiler will declare Foo() for you
};

Now Foo() is automatically declared; but it's not necessarily defined. If class Foo has a reference member (e.g. int& a), Foo obviously can't be default constructed (there's a longer list of rules here, if you are in the mood of reading legal-ish documents). Other examples include e.g. what if there's one member that can't be default constructed. In these cases, the compiler will delete the default constructor – same as Foo() = delete;. If no issues are found, the compiler will define a default constructor for you that is equivalent to an empty function.

This is nice, especially for a templated class that you can't determine ahead of time if T can be default constructed or not. This is cool! But what if you defined one constructor? You will loose the implicitly declared constructor from the compiler! That's what Foo() = default; is for. You can enforce the compiler to perform the aforementioned implicit thingie while having your own constructors.