float64 is Go’s double-precision floating-point type. It stores a finite approximation of real values using 64 bits: 11 exponent bits, 52 mantissa bits, and one sign bit. Floating-point arithmetic therefore may lose precision. section 3.3
Mistake: expecting exact decimal arithmetic
Multiplying 1.0001 by itself may print 1.0002 instead of the expected 1.00020001. The stored value cannot represent every real number exactly, so arithmetic operates on approximations. section 3.3
var n float32 = 1.0001 fmt.Println(n * n)
Running a floating-point multiplication exposes the difference between expected decimal arithmetic and the value produced by approximation. section 3.3
Account for these behaviors when using float64:
- Changing operation order can change accumulated error: adding 1.0001 repeatedly before adding 10,000 produces different results from adding 10,000 first. section 3.3
- Algebraically equivalent expressions can differ numerically: a × (b + c) and a × b + a × c produce slightly different outputs. section 3.3
- Floating-point values can represent positive infinity, negative infinity, and NaN through operations such as division by zero. section 3.3
Mistake: assuming decoded JSON numbers are integers
Unmarshaling JSON into map[string]any converts numeric values to float64, even when the JSON value has no decimal. Code that assumes m["id"] is an integer can therefore make an incorrect type assertion and cause problems. section 10.3.3
fmt.Printf("%T\n", m["id"])
float64
Printing the dynamic type of a decoded numeric map value reveals float64 directly. section 10.3.3
