Double comparison — epsilon vs BigDecimal for money?

Question

Accepted Answer

Comparing doubles double and float use IEEE 754 binary fractions—many decimal values (0.1, money) are not exact. Never use == for business rules; use epsilon tolerance for physics, BigDecimal for money. Use case Approach Avoid Money BigDecimal with scale + compareTo double cents Sensor math epsilon compare raw == Sorting Double.compare subtract and compare to zero BigDecimal price = new BigDecimal("19.99"); BigDecimal tax = price.multiply(new BigDecimal("0.0825")); if (price.compareTo(tax) < 0) { /* ... */ } // Not for money: double bad = 0.1 + 0.2; // 0.30000000000000004 Say "store money as integer minor units or BigDecimal strings, never binary float."

Use case	Approach	Avoid
Money	`BigDecimal` with scale + `compareTo`	`double` cents
Sensor math	epsilon compare	raw `==`
Sorting	`Double.compare`	subtract and compare to zero