I see a couple of problems in there.

@robpike could you elaborate?

Hm... fmt documentation states that:
Flags are ignored by verbs that do not expect them. For example there is no alternate decimal format, so %#d and %d behave identically.
But go vet reports unrecognized flags as error. It this even correct? I understand that removing such redundancies from a source code is a good thing. But what about go vet executed before tests in Go 1.10?

Speaking about other problems of '0' flag processing by go vet.

fmt.Printf("%020p \n", &[]int{})
fmt.Printf("%020T \n", false)
fmt.Printf("%020U \n", 1234)

In all these cases '0' affects print result, but is reported by go vet as unrecognized.

This does not seem like unreasonable go vet behavior to me. The code is almost certainly mistaken even if fmt carries on as best it can.

Here is an additional test case using fmt.Formatter to let you decide what should be the bahaviour of "go vet":

package main

import (
	"fmt"
)

type t struct{}

func (_ t) Format(f fmt.State, c rune) {
	if f.Flag('0') {
		f.Write([]byte("GA"))
	} else {
		f.Write([]byte("BU"))
	}
}

func main() {
	fmt.Printf("%0s\n", t{})
}

https://go-review.googlesource.com/c/go/+/90695 was supposed to mention this issue but got the issue number wrong.

Fixed by https://golang.org/cl/90695.