Question

A std::array<T> is essentially a C-style array wrapped in a struct. The initialization of structs requires braces, and the initialization of arrays requires braces as well. So I need two pairs of braces:

std::array<int, 5> a = {{1, 2, 3, 4, 5}};

But most of the example code I have seen only uses one pair of braces:

std::array<int, 5> b = {1, 2, 3, 4, 5};

How come this is allowed, and does it have any benefits or drawbacks compared to the first approch?

Answer 1

The benefit is that you have … less to type. But the drawback is that you are only allowed to leave off braces when the declaration has that form. If you leave off the =, or if the array is a member and you initialize it with member{{1, 2, 3, 4, 5}}, you cannot only pass one pair of braces.

This is because there were worries of possible overload ambiguities when braces are passed to functions, as in f({{1, 2, 3, 4, 5}}). But it caused some discussion and an issue report has been generated.

Essentially, the = { ... } initialization always has been able to omit braces, as in

int a[][2] = { 1, 2, 3, 4 };

That’s not new. What is new is that you can omit the =, but then you must specify all braces

int a[][2]{ {1, 2}, {3, 4} };