The auto keyword changed its semantics starting from the C++11 standard. In this notebook, we will review its new semantics and examine its new flavors.

Using the auto keyword

The auto keyword changed its semantics starting from the C++11 standard. From C++11 onwards, the semantic of the keyword is automatic type deduction. In fact, from C++11 onwards, we have the following different flavors of the keyword [1]:

• auto
• const auto&
• auto&
• auto&&

Furthermore, we have decltype(auto). This post is a short guide on how to use auto and its various flavors.

The first thing to note about auto is that it ise used for automatic type deduction. This means that we can write code like the following:

...
// before c++11
int x = 5;

//from c++11
auto x = 5;
...

Thus, using auto may help us to write cleaner and less cluttered code. This is emphasized in particular when we consider function signatures. Compare the code snippet below (example taken from [1]):

class Foo
{

int value()const{..}
const int& cref_value()const{...}
int& ref_value(){...}

};

with the following code snippet

class Foo
{

auto value()const{..}
auto& cref_value()const{...}
auto& ref_value(){...}

};

The latter API is cleaner and simpler.The type returned is deduced by the compiler for us.

Remark

Althgough to a large extent using auto simplifies our code, overusing it can have the opposite result.

One other advantage of using auto is that we cannot leave the variable uninitialized. That is the following fails to compile

...
auto x;
...

This is reasonable as the compiler uses the right hand side value to deduce the type and hence the memory size it has to allocate. If there isn’t a value there is nothing to deduce from. Thus, unitialized variables are not allowed when using auto and the good news are that the compilers let us know the exact line number in our code that this occurs.

References

1. Bjorn Andrist, Viktor Sehr, C++ High Performance, 2nd Edition, Packt Publishing