},
		{
			name: "max",
			fields: fields{
				value: []byte(strconv.FormatFloat(math.MaxFloat64, 'g', -1, 64)),
			},
			want:   math.MaxFloat64,
			wantOK: true,
		},
		{
			name: "min",
			fields: fields{
				value: []byte(strconv.FormatFloat(-math.MaxFloat64, 'g', -1, 64)),
			},
			want:   -math.MaxFloat64,
			wantOK: true,
		},
		{
			name: "max-overflow",
			fields: fields{

// swap swaps the elements with indexes i and j.
func Shuffle(n int, swap func(i, j int)) { globalRand.Shuffle(n, swap) }

// NormFloat64 returns a normally distributed float64 in the range
// [-math.MaxFloat64, +math.MaxFloat64] with
// standard normal distribution (mean = 0, stddev = 1)
// from the default Source.
// To produce a different normal distribution, callers can
// adjust the output using:
//

func Read(p []byte) (n int, err error) { return globalRand().Read(p) }

// NormFloat64 returns a normally distributed float64 in the range
// [-[math.MaxFloat64], +[math.MaxFloat64]] with
// standard normal distribution (mean = 0, stddev = 1)
// from the default [Source].
// To produce a different normal distribution, callers can
// adjust the output using:
//

 * (Marsaglia & Tsang, 2000)
 * https://www.jstatsoft.org/v05/i08/paper [pdf]
 */

const (
	re = 7.69711747013104972
)

// ExpFloat64 returns an exponentially distributed float64 in the range
// (0, +math.MaxFloat64] with an exponential distribution whose rate parameter
// (lambda) is 1 and whose mean is 1/lambda (1).
// To produce a distribution with a different rate parameter,
// callers can adjust the output using:
//

 * (Marsaglia & Tsang, 2000)
 * https://www.jstatsoft.org/v05/i08/paper [pdf]
 */

const (
	re = 7.69711747013104972
)

// ExpFloat64 returns an exponentially distributed float64 in the range
// (0, +[math.MaxFloat64]] with an exponential distribution whose rate parameter
// (lambda) is 1 and whose mean is 1/lambda (1).
// To produce a distribution with a different rate parameter,
// callers can adjust the output using:
//

	if rand.Int()&1 == 1 {
		f = -f
	}
	return float32(f)
}

// randFloat64 generates a random float taking the full range of a float64.
func randFloat64(rand *rand.Rand) float64 {
	f := rand.Float64() * math.MaxFloat64
	if rand.Int()&1 == 1 {
		f = -f
	}
	return f
}

// randInt64 returns a random int64.
func randInt64(rand *rand.Rand) int64 {
	return int64(rand.Uint64())
}

	"1/3",
}

// isFinite reports whether f represents a finite rational value.
// It is equivalent to !math.IsNan(f) && !math.IsInf(f, 0).
func isFinite(f float64) bool {
	return math.Abs(f) <= math.MaxFloat64
}

func TestFloat32SpecialCases(t *testing.T) {
	for _, input := range float64inputs {
		if strings.HasPrefix(input, "long:") {
			if !*long {
				continue
			}
			input = input[len("long:"):]

		// Iterate through the various n-choose-k NUMA node combinations,
		// looking for the combination of NUMA nodes that can best have CPUs
		// distributed across them.
		var bestBalance float64 = math.MaxFloat64
		var bestRemainder []int = nil
		var bestCombo []int = nil
		acc.iterateCombinations(numas, k, func(combo []int) LoopControl {
			// If we've already found a combo with a balance of 0 in a

pkg math (windows-amd64), const MaxInt = 9223372036854775807
pkg math (windows-amd64), const MaxUint = 18446744073709551615
pkg math (windows-amd64), const MinInt = -9223372036854775808

			want:          float64(math.SmallestNonzeroFloat64),
			assertOnError: assertNilError,
		},
		{
			name:          "max float64 value",
			in:            hex("fb7fefffffffffffff"),
			want:          float64(math.MaxFloat64),
			assertOnError: assertNilError,
		},
		{
			name:          "max float32 value as double precision",
			in:            hex("fb47efffffe0000000"),
			want:          float64(math.MaxFloat32),