为什么也将C和C ++关键字“ #defined”或“ typedefed”作为其大写版本？

Question

Sorry for the confusing title but I couldn't think of a title that would describe my question clearly. When writing a C or C++ program, you can usually refer to keywords such as char with CHAR, or int with INT, or void with VOID, and on.

In winnt.h, they are defined as follows:

#ifndef VOID
#define VOID void
typedef char CHAR;
typedef short SHORT;
typedef long LONG;
#if !defined(MIDL_PASS)
typedef int INT;
#endif
#endif

What is the use of doing this? Is it to satisfy convention or ensure compatibility?

Answer 1

I've always thought of these trivial typedefs (eg int -> INT) to be more about consistency of than anything.

What I mean by this, is that there are many many typedefs in Win32 in caps - DWORD, WORD, LPARAM, WPARAM, HANDLE, HWND, some of these are simply integers and their name provides context as to their use, and some of them (eg RECT and WNDCLASS) represent structures.

Consider the case of function void f(int, BOOL, char, PAINTSTRUCT); to me it feels more consitent to allow all capitals: VOID f(INT, BOOL, CHAR, PAINTSTRUCT). But that could just be me.

Answer 2

Mostly convention if I'm not mistaken. I think windows put some of them in their to ensure compatibility for some data types (integers for example, maintaining compatibility of 8 bit, 16 bit, 32 bit, etc.).

Answer 3

The capitalized (#defined) versions are often used to allow the code to be made compatible with various compilers. And typedefs are typically used to ensure the newly named types meet stricter rules than the originals defined by the compiler.

Typically, typedef int INT , or the like can be seen inside blocks that are detecting the compiler-environment, so that a name like "INT" can be used to provide more restrictive semantics that "int".

In "C", all primitive types, like int and char , are "at least" a certain size (I don't know that size right now). Which means the compiler is allowed to make them larger. I know of a system that used a 32-bit char type. This leads to easily writing non-portable code when assumptions are made, such as assuming "int" is always 32 bits.

An experiment in your environment might help. Try a small program like the following and then compile and run it with 32-bit settings and then 64-bit settings:

int main (int argc, char **argv)
{
    printf("sizeof INT=%u\n", sizeof (INT));
    printf("sizeof int=%u\n", sizeof (int));
}

As far as VOID, defining it to "void" is a little odd. However, it looks like - in the snippet given - it's being used to determine if the others have already been defined by the same header file - which is a very important step in "C" since duplicate typedefs lead to errors. IIRC, duplicate defines can lead to errors or warnings, denpending on the compiler and settings.