Convert int64_t to double

Question

  int64_t a = 1234;

  double d = (double) a;

Is this the recommended way?

Answer 1

You should use static_cast or rely on the implicit cast instead:

int64_t a = 1234;
double d = static_cast<double>(a);
double f = a;

Answer 2

use static_cast as strager answers. I recommend against using the implicit cast (or even a C-style cast in C++ source code) for a few reasons:

• Implicit casts are a common source of compiler warnings, meaning you may be adding noise to the build (either now, or later when better warning flags are added).
• The next maintenance programmer behind you will see an implicit cast, and needs to know if it was intentional behavior or a mistake/bug. Having that static_cast makes your intent immediately obvious.
• static_cast and the other C++-style casts are easy for grep to handle.

Answer 3

For POD types both versions do the same thing. Choose the one you prefer and be consistent.

I know many people who prefer the former for typing/readability and I tend to agree with this but I can live with either.

I've heard the "easy to grep for" argument many times but have yet to ever come across a situation where I've needed to grep my codebase for POD casts.

Answer 4

You can also use the conversion syntax, which is equivalent to a static_cast:

int64_t a = 1234;
double d = double(a);

This is a useful syntactic construction in that it allows primitive and class types to be treated the same way in template code, either doing a static_cast, for the primitive, or invoking a constructor, for the class type.