I'm curious about whether it's best to use an int or an unsigned 1 byte type for representing booleans in C, especially if I'm working with older standards prior to C23. I know that bool is officially introduced in C23, but what about earlier versions? Are there any performance implications I should consider?
3 Answers
I lean towards using int too, mainly because of memory alignment and speed. It usually performs better, especially on 64-bit systems. Using smaller types like a single byte might not provide the same performance benefits.
In C23, bool is officially available, but if you're sticking to an older standard, both unsigned int and int work fine for representing boolean values. In fact, you can even use other types like char, long, or short if you want, but generally, int is a safe bet as it aligns well in memory and offers good performance.
I'd suggest going with int. It works seamlessly with all comparison operators and conditional statements. Plus, C traditionally uses int for boolean logic, so you might find it easier to read and maintain your code. I only recommend a 1 byte type if you're in a really tight memory space or if you're packing booleans into structures. Otherwise, you might run into some awkward alignment issues if you're not careful.
Back when I had 640K RAM, I used unsigned integers to squeeze in 16 booleans into 2 bytes, but it required a lot of bit manipulation to access them. Definitely a trade-off!
Why would packing booleans in a struct cause different issues? Is it related to memory alignment?