Wrapping arithmetic

# Wrapping arithmetic (decided 2026-09-20): plus.wrap minus.wrap times.wrap pow.wrap are always defined — modulo the
# type's size — so a hash needs no no_check. Plain plus minus times pow stay overflow-checked.
fn hash [lent 'text': str.utf8] -> uint32 {
    let 'h': uint32 = 2_166_136_261;                 # FNV-1a
    for 'b' in str.each.byte[lend 'text'] {
        change 'h' = 'h' xor 'b';                          # FNV-1a: xor, then a wrapping multiply
        change 'h' = 'h' times.wrap 16_777_619;                    # wraps instead of overflowing
    }
    return 'h';
}

MAIN {
    std::print.useful[hash["hello"], newln];
    let 'c': uint8 = 250;
    change 'c' = 'c' plus.wrap 10;                                # 260 wraps to 4
    std::print.useful['c', newln];
    let 'i': int8 = 127;
    change 'i' = 'i' plus.wrap 1;                                 # 128 wraps to -128
    std::print.useful['i', newln];
    let 'p': uint16 = 3;
    change 'p' = 'p' pow.wrap 20;                               # 3^20 modulo 65536
    std::print.useful['p', newln];
}