errors.Is() behavior with oops errors

[thejoeejoee]

Hi! At first, thanks for this great package, really simplifies error building.

I've noticed some weird behavior when using errors.Is with oops errors.
Currently, oopsError.Is(oopsError) returns false for the same error instance, which is counter-intuitive. I would expect that any error should be considered to be "Is" itself, similar to how the standard errors package works.

Demonstration test case:

package main

import (
	"errors"
	"testing"

	"github.com/samber/oops"
	"github.com/stretchr/testify/assert"
)

var MyErr1 = errors.New("my custom error")

var MyOopsErr = oops.New("my oops error")

func TestErrIs(t *testing.T) {
	// is fine
	assert.ErrorIs(t, MyErr1, MyErr1, "Standard error should be Is itself")

	// fails
	identity := errors.Is(MyOopsErr, MyOopsErr)
	assert.True(t, identity, "Oops error should be Is itself")

	// also fails
	oopsed, _ := oops.AsOops(MyOopsErr)
	assert.True(t, oopsed.Is(MyOopsErr), "Converted Oops error should be Is itself")

	// fails
	assert.ErrorIs(t, MyOopsErr, MyOopsErr, "errors.Is should work with Oops errors")
}

It's a bit inconvenient especially during tests, where testify/assert.ErrorIs is reporting a failure for oops errors.

Is this intentional behavior?

[samber]

Thanks @thejoeejoee for the report. We made a fix recently on that method but a unit test was missing.

Please upgrade to v1.19.4