int24_t, char24_t with exactly 3 bytes

[E86M92yv]

Hello,
sizeof(wchar_t) is 2. It can store any character from 0 - 65535. That is why I need an universal integer which can store a Unicode character at any range. 2^24 = 16777216 should cover them all. char32_t is just too much and it wastes lots of spaces. I have tried, but suprisingly the size of char24_t is still 4 bytes not 3 bytes.

#define char24_t int24_t
struct int24_t
{
    int32_t val : 24;
};

Could anyone help me solve the problem? Thanks.

[JLBorges]

Create a user defined type char24_t which holds three bytes, and has implict conversions to/from char32_t.

Something along these lines:

#include <iostream>
#include <limits>
#include <iterator>
#include <algorithm>

struct char24_t
{
    static constexpr std::size_t NBITS = std::numeric_limits<unsigned char>::digits ;
    static_assert( NBITS == 8, "byte must be an octet" ) ;

    char24_t() = default ;
    char24_t( char32_t value )
    {
        // assert value within range of char24_t
        lsb = value & 0xff ;
        value >>= NBITS ;
        midb = value & 0xff ;
        msb = value >> NBITS ;
    }

    operator char32_t() const { return ( msb << (NBITS*2) ) + ( midb << NBITS ) + lsb ; }

    std::uint8_t msb = 0 ;
    std::uint8_t midb = 0 ;
    std::uint8_t lsb = 0 ;
};

int main()
{
    constexpr std::size_t N = 100 ;
    const char32_t a32[N] = { 0, 111, 2222, 33333, 444444, 5555555, 16777215 } ;
    char24_t a24[N] ;
    std::cout << "sizeof(a32): " << sizeof(a32) << "    sizeof(a24): " << sizeof(a24) << '\n' ;

    std::copy( std::begin(a32), std::end(a32), std::begin(a24) ) ;
    std::cout << std::boolalpha << "equal? " << std::equal( std::begin(a32), std::end(a32), std::begin(a24) ) << '\n' ;
}

http://coliru.stacked-crooked.com/a/184655a3b9cfab7b