Supporting Multiple Platforms in C++ - Learn C++ for Game Development - page 260

The int8_t and uint8_t provide integers that are 8 bits or one byte in length. The u version is
unsigned and the non-u version is signed . The other types are similar but of their equivalent fixed
length. There are 16-bit versions, 32-bit versions, and 64-bit versions of integers.
Note You should avoid using the 64-bit integers for the time being unless you explicitly need numbers that
cannot be stored within 32 bits. Most processors still operator on 32-bit integers when doing arithmetic. Even
64-bit processors that have 64-bit memory addresses for pointers still do normal arithmetic using 32-bit ints.
In addition, 64-bit values use twice as much memory as 32-bit values, which increases the RAM required to
execute your program.
The next problem that might arise is that the char type might not be the same on all platforms.
C++ does not supply a fixed-size char type, so we need to improvise a little. Every platform I have
developed games on has used 8-bit char types, so we're only going to account for that. We will,
however, define our own char type alias so that if you ever do port code to a platform with chars
larger than 8 bits then you will only have to solve the problem in a single place. Listing 26-1 shows
the code for a new header, FixedTypes.h .
Listing 26-1. FixedTypes.h
#pragma once
#include <cassert>
#include <cstdint>
#include <climits>
static_assert(CHAR_BIT == 8, "Compiling on a platform with large char type!");
using char8_t = char;
using uchar8_t = unsigned char;
The FixedTypes.h file includes cstdint , which gives us access to the 8- to 64-bit fixed-width
integers. We then have a static_assert that ensures that the CHAR_BIT constant is equal to 8.
The CHAR_BIT constant is supplied by the climits header and contains the number of bits that are
used by the char type on your target platform. This static_assert will ensure that our code, which
includes the FixedTypes header, will not compile on platforms that use a char with more than 8
bits. The header then defines two type aliases, char8_t and uchar8_t , which you should use when
you know you specifically need 8-bit chars . This isn't necessarily everywhere. Generally you will
need 8-bit char types when loading data that was written out using tools on another computer that
did use 8-bit character values because the length of the strings in the data will have one byte per
character rather than more. If you're not sure if you do or don't need 8 bits specifically, you're better
sticking to always using 8-bit chars .
The last problem solved in the cstdint header is for using pointers on platforms with different-sized
pointers to integers. Consider the code in Listing 26-2.