Hacker News Reader

> Nothing - nothing - needs more than about four or five significant digits.

So if you have a million in the bank, you would be fine with a 100$ rounding error on each transaction?

You don't use floats for money, because it's an integer.

Oddly, money is not an integer at large scales. At least, if you meant to calculate it in pennies, that doesn't work. Consider, what is the cost of gas?

I'm not quite sure what you mean? In most any case I can imagine, uint64 is quite enough to count money in pennies, assuming inflation doesn't get truly out of hand.

Gas is priced in sub penny values. Similarly, returns and other values are all typically calculated in sub penny values.

You could store in 1/100th of a penny, I suppose? I'm not entirely sure that will work out for all calculations.

At present? Price at the pump is 72.9 pence per litre, which is represented internally in the software as 7290 1/100ths of a penny.

Right, my point was just that "use pennies" won't work. You have to use fractions of a penny for different tasks. I'm not sure on the implications of using different fractions at different levels, but I'm sure the problem is more complicated than it sounds. Even before you get into the various methods of rounding that exist.

Reply on news.ycombinator.com