On: Systemd is bad parsing

[viraptor]

> the C standard requires that sizeof char be exactly one byte

That's at least 1, not exactly 1.

It's POSIX that requires char to be 8bit.

[evincarofautumn → viraptor]

It’s “exactly one byte” in the sense that by definition, sizeof(char) = 1, but it’s “at least one byte” in the sense that CHAR_BIT ≥ 8.

[alxlaz → viraptor]

Nope, the C standard requires that sizeof char = 1, regardless of the width of a byte :-).

[jstimpfle → alxlaz]

One of the things I was not sure about is whether the size of uint8_t is guaranteed to be 1 as well. And whether that means the both are pointer-compatible. And if so, where is this specified?

[alxlaz → jstimpfle]

As far as I know, C99 makes no requirements about the byte width of int8_t, only about its bit width. The relevant bits are in section 7.18.1.1 .

There are few architectures and runtimes where this is relevant anymore, and the answer is really best given on a case-by-case basis.

I think it's technically possible to make the two pointer-compatible on architectures with wider bytes. Last time I used one it was Analog Devices' SHARC, and I think they were pointer-compatible, but that was a long time ago, so take it with a grain of salt.