4 replies to this topic

[hkBattousai]

I have this constructor of the template class PolynomialFraction:

template <class T>
PolynomialFraction<T>::PolynomialFraction(uint64_t unNumNomCoef, uint64_t unNumDenCoef, T FirstCoefficient, ... )
{
	std::vector<T> NominatorCoefficients;
	std::vector<T> DenominatorCoefficients;
	va_list ArgumentPointer;
	va_start(ArgumentPointer, FirstCoefficient);
	T NextCoefficient = FirstCoefficient;
	if (unNumNomCoef == 0) // Every polynomial must have at least one coefficient
	{
		NominatorCoefficients.push_back(static_cast<T>(0));
	}
	else
	{
		NominatorCoefficients.push_back(NextCoefficient);
		for (uint64_t i=1; i<unNumNomCoef; i++)
		{
			NextCoefficient = va_arg(ArgumentPointer, T);
			NominatorCoefficients.push_back(NextCoefficient);
		}
	}
	if (unNumNomCoef == 0)
	{
		DenominatorCoefficients.push_back(static_cast<T>(1));
	}
	else
	{
		for (uint64_t i=0; i<unNumDenCoef; i++)
		{
			NextCoefficient = va_arg(ArgumentPointer, T);
			DenominatorCoefficients.push_back(NextCoefficient);
		}
	}
	m_NominatorPolynomial = Polynomial<long double>(NominatorCoefficients);
	m_DenominatorPolynomial = Polynomial<long double>(DenominatorCoefficients);
}

As its name suggests, this class is supposed to hold nominator and denominator polynomials of a polynomial fraction.

Constructor parameters are:
First: Number of coefficients in nominator
Second: Number of coefficients in denominator
Third: First coefficient
The rest: The other coefficients if any.

I'm calling this constructor like below:

PolynomialFraction<long double> pf1(2, 3, 1, 2, 1, 2, 3);
std::cout << pf1.ToString() << std::endl;

Its output is:

(-9.25596e+061x + 1)/(-9.25596e+061x^2 - 9.25596e+061x - 9.25596e+061)

(It is not a bug in ToString() method, I checked it.)

But it should have been:

(2x + 1)/(3x^2 + 2x + 1)

In other words, I don't receive any coefficient except for the very first one.

However, the code below which is a very similar one is working flawlessly:

template <class T>
Polynomial<T>::Polynomial(uint64_t NumberOfCoefficients, T FirstCoefficient, ...)
{
	if (NumberOfCoefficients == 0)
	{
		m_Coefficients.push_back(static_cast<T>(0));
		return;
	}
	va_list ArgumentPointer;
	va_start(ArgumentPointer, FirstCoefficient);
	T NextCoefficient = FirstCoefficient;
	std::vector<T> Coefficients;
	Coefficients.push_back(NextCoefficient);
	for (uint64_t i=1; i<NumberOfCoefficients; i++)
	{
		NextCoefficient = va_arg(ArgumentPointer, T);
		Coefficients.push_back(NextCoefficient);
	}
	m_Coefficients = Coefficients;
}

What am I doing wrong in my code?

[Josh Petrie]

Your parameters are integers, not doubles. The compiler cannot tell that the arguments passed via the ellipsis should undergo conversion; try passing them explicitly as double literals (that is, 1.0, 2.0, 3.0, et cetera).

[Martins Mozeiko]

You are passing integers to PolynomialFraction constructor (2, 3, 1, ...), but you are expecting long double (second T argument to va_arg macro). That won't work. Pass the correct type.

[hkBattousai]

Ohh guys, thank you.
It is working now.

[Washu]

and the reason you're not just simply passing in an array, or an std::vector<double> directly? Instead of playing around with the type-unsafe and quite undefined behavior prone ellipses operator?