What is the correct format specifier to print error object in Go: %s or %v?

Question

Here is my program.

package main

import (
    "errors"
    "fmt"
)

func main() {
    a := -1
    err := assertPositive(a)
    fmt.Printf("error: %s; int: %d\n", err, a)
    fmt.Printf("error: %v; int: %d\n", err, a)
}

func assertPositive(a int) error {
    if a <= 0 {
        return errors.New("Assertion failure")
    }
    return nil
}

Here is the output.

error: Assertion failure; int: -1
error: Assertion failure; int: -1

In this program, it makes no difference whether I use %s or %v to print the error object.

I have two questions.

1. Is there any scenario while printing errors where it would make a difference for %s and %v?
2. What is the correct format specifier to use in this case?

Solution

According to docs:

%v  the value in a default format
...
%s  the uninterpreted bytes of the string or slice

Also, more information about error:

The error type is an interface type. An error variable represents any value that can describe itself as a string.

So, treat it as %s.

Comments

[Volker]

Use %s (but please look up what %v and %s do in the package description of fmt).

[Lone Learner]

@Volker I know what %v and %s does. I would like to understand why you recommend %s for error objects. Can you share the rationale please?

[Volker]

Less indirection. What gets printed is the output of the Error() method and that's a string: Clearer.

[Nathan Lutterman]

The only thing that I'd say is perhaps confusing about this is that the Golang team uses %v in their blogs: blog.golang.org/error-handling-and-go. Otherwise I think it's probably okay to just use %s as well.

[Nathan Lutterman]

Also more discussion in this issue here: github.com/golang/go/issues/29934, which is somewhat relevant

[Jeff Learman]

Is it worth mentioning the special handling for format strings ending in ": %s" followed by an error? That's very non-intuitive (if it was retained in the language; I'm not sure.) See @NathanLutterman's citation.