math: MaxUint64 doesn't return correct result but an error?

[IIvyPy]

What version of Go are you using (go version)?

$ go@1.12

Does this issue reproduce with the latest release?

What operating system and processor architecture are you using (go env)?

$ amd64

What did you do?

I just use the const in math packge: math.MaxUint64

What did you expect to see?

18446744073709551615

What did you see instead?

./main.go:169:29: constant 18446744073709551615 overflows uint

Notice that "1<<64" in "math.MaxUint64 = 1<<64 - 1" may cause overflow, perhas you can use "math.MaxUint64 = ^uint64(0)" instead.

[randall77]

Show us your code. This code works fine:

package main

import "math"

func main() {
	var x uint = math.MaxUint64
	_ = x
}

The definition of MaxUint64 as 1<<64-1 has no issues with overflow. It is a constant expression, which in Go is evaluated with arbitrary precision/range.

[IIvyPy]

Show us your code. This code works fine:

package main

import "math"

func main() {
	var x uint = math.MaxUint64
	_ = x
}

The definition of MaxUint64 as 1<<64-1 has no issues with overflow. It is a constant expression, which in Go is evaluated with arbitrary precision/range.

Your code works well.
Here is my code.

package main

import "math"
func main() {
 	x := math.MaxUint64
	_ = x
 }

[zhangfannie]

Is this related to untyped constant? The untyped constant math.MaxUint64 cannot infer the type from x that has no type. So it is given the default type int, so it overflows.

[bmkessler]

Yes, the second example is due to the type inference with the untyped constant.

x := math.MaxUint64

infers that x is an int and it overflows.

Also, the first example

var x uint = math.MaxUint64

still overflows on any architecture where uint is 32-bit, e.g. arm.

If you want to assign math.MaxUint64 to a variable you need to store it in a uint64

[randall77]

Closing, this is working as intended.
As @bmkessler says, explicitly type your variable instead of using :=.

[ianlancetaylor]

See also #19621, #23086, #23087, #23163.