hex_print

hex_print.h

#include <cstdio>
#include <cstdint>

template <typename T>
void hex_print(T v) {
	switch(sizeof(T)) {
		case  1: printf("0x%01X\n", *(reinterpret_cast<uint8_t  *>(&v))); break;
		case  2: printf("0x%02X\n", *(reinterpret_cast<uint16_t *>(&v))); break;
		case  4: printf("0x%04X\n", *(reinterpret_cast<uint32_t *>(&v))); break;
		case  8: printf("0x%08llX\n", *(reinterpret_cast<uint64_t *>(&v))); break;
		case 16: printf("0x%08llX%08llX\n", *(reinterpret_cast<uint64_t *>(&v) + 1), *(reinterpret_cast<uint64_t *>(&v))); break; // for long double (but this is incorrect...!!)
		default: break;
	}
}

test.cpp

#include <iostream>
#include "hex_print.h"

using namespace std;
int main(int argc, char *argv[]) {
	hex_print('B');
	hex_print(4u);
	hex_print(9.0f);
	hex_print(9.2);
	hex_print(9.9l);
	hex_print(9.8l);
	
	return EXIT_SUCCESS;
}

So the default for an unspecified floating point number (92) is uint64_t / native word size? And if you specify f is always uint32_t? Interesting that specifying f will actually reduce accuracy (at least on 64-bit systems).

Such a great exploration. I will continue writing about it when I get back from Kansai :)

@jedahan
oh... sorry, i find your comment now...

1.0 is typed double on every C/C++/Obj-C enviroment.
and, definition of double is 64bit in IEEE 754. therefore, convert to uint64_t.

if we wanna use float definitely, we have to add suffix f.
then, always it is typed float and convert to uint32_t.

so, i like adding suffix f when writing oF codes. because oF use float basically.
(but, most people does not seem to mind... 😿 )

omg,

        case 16: printf("0x%08llX%08llX\n", *(reinterpret_cast<uint64_t *>(&v) + 1), *(reinterpret_cast<uint64_t *>(&v))); break;

is wrong...