On Sunday, March 30, 2025, Thiago Macieira wrote:

> On Saturday, 29 March 2025 17:04:46 Eastern Daylight Time Jonathan Wakely
> wrote:

> On what platform in existence today does a type that's at least 32 bits
> > promote?
>
> None. It would need to be a platform with an int probably of 64 bits,
> which
> doesn't exist.
>


I'm currently writing a fully-portable (C++14) implementation of MD5
constexpr, in order to make a single-header-only library to generate UUID's
at compile time. I'm testing it on 27 cross-compilers, both big endian and
little endian, as well as on my trusty Texas Instruments microcontroller
that has a 16-Bit char and a 16-Bit int.

CHAR_BIT == 16 && 1==sizeof(short)==sizeof(int)

Also I'm accommodating one's complement and sign magnitude (even though
C++20 has outlawed them).

Another thing I'm accommodating is:

CHAR_BIT==64 && sizeof(short)==sizeof(long long)

So anyway my code looks a little ridiculous in places as I'm super-paranoid
about stuff such as a 32-Bit unsigned int promoting to a 64-Bit signed int.

I wrote an article 15 years ago about integer promotion . . . if you scroll
down to the very bottom you'll see what happens if you invert the bits of
an unsigned 16-Bit integer on a one's complement machine:

http://www.virjacode.com/tutorials/c_int_promotion/index.html