Question

I’m experimenting with templates and I wrote this simple class method:

void Decimal::toBinary(size_t bits) {
    // decimalNumber being a class private variable (long double)
    std::bitset< bits > result(decimalNumber);

    std::cout << result << std::endl;
}

I’m trying to pass size_t bits function argument to the bitset template.

According to C++ Bitset Reference the implementations takes indeed a size_t argument:

template < size_t N > class bitset;

However, I’m getting

src/decimal.cc:11: error: ‘bits’ cannot appear in a constant-expression
src/decimal.cc:11: error: template argument 1 is invalid
src/decimal.cc:11: error: invalid type in declaration before ‘(’ token

I suppose I’m not able to do this… any workarounds?

Answer 1

Templates are a compile-time feature, not a run-time feature. If the number of different sizes you are going to support is limited, e.g., because you support 8, 16, and 32 bits, you can switch and delegate:

template <size_t bits>
void Decimal::toBinary() {
    std::bitset< bits > result(decimalNumber);
    std::cout << result << '\n';
}
void Decimal::toBinary(size_t bits) {
    switch (bits) {
    default: throw std::runtime_error("number of bits not supprted");
    case  8: this->toBinary< 8>(); break;
    case 16: this->toBinary<16>(); break;
    case 32: this->toBinary<32>(); break;
    }
}

BTW, don’t use std::endl. If you really mean to flush the stream use std::flush.