maemo.org - Talk - View Single Post - Floating point questions in C++...

sony123	2013-01-13 , 22:55
Posts: 650 \| Thanked: 619 times \| Joined on Nov 2009	#1

I have floating point computation in Qt C++, and I am having precision issues. The function is called in Javascript and the argument is floating point.

Code:

Q_INVOKABLE QVariantList FileHandler::calcUniSymPerf (const int i, const float price) {
...
}

Let's say the price argument is 34.82, and when I use qDebug() to print out price variable it is 34.82, but converting it to QString using QString::number(price, 'f', 8) gives me 34.81999969. I try to round the price to two decimal points:

Code:

    
// Code 1
    float nearestPrice = (floor(price * 100 + 0.005)); 
    nearestPrice = nearestPrice/100;
// Code 2
    float nearestPrice = (floor(price * 100 + 0.005))/100.0;
    qDebug() << "nearest=" << QString::number(nearestPrice,'f',8);

The results for code 1 and code 2 are different. Code 1 gives me the 34.82000000 I wanted, but code 2 gives me 34.81999969.
Further down the code, I want to calculate the market value by mutiplying price and share (QVariant converted to float).

Code:

    float value   = price * uniShare.at(i).toFloat();

The share is 30005, so the correct value is 1044774.10, but the program gives me 1044774.06.Even multiplying 34.81999969*30005 should be 1044774.09.

I am at a lost. Can anyone educate me on why the floating point calculations are not working?

Quote & Reply |