Don't panic on nil callback?

[abuchanan-nr]

What if instead of panicing on a nil ...Func value, you just return the zero values?

Sometimes our tests don't need to return anything, we just want to capture and verify the calls. The mock.calls data structure is great for that, but adding empty implementations of ...Func adds boilerplate.

[Mat Ryer]

The question is whether people will be surprised by what's going on or not. What do you think?

[Lucas Bremgartner]

I think it is an idea worth exploring, but if added to moq, it should be an opt-in feature which does not change the current default behavior.

See also #36

[Mat Ryer]

I would be open to a PR that added this as a flag. -defaults or similar.

[Lucas Bremgartner]

I have a situation, where this feature would be beneficial for me, so I am interested in having this.

See also the (quite old) PR #96 for a potential implementation.

[natan streppel]

Came here looking for this. As far as I can see this hasn't been implemented yet, right?

[Suhas Karanth]

Moq now supports a -stub flag to support this use-case. Example of generated code when flag is specified:

moq/pkg/moq/testpackages/channels/queuer_moq.golden.go

Lines 57 to 74 in 83ab8dd

 func (mock *QueuerMock) Sub(topic string) (<-chan Queue, error) { 
 	callInfo := struct { 
 		Topic string 
 	}{ 
 		Topic: topic, 
 	} 
 	mock.lockSub.Lock() 
 	mock.calls.Sub = append(mock.calls.Sub, callInfo) 
 	mock.lockSub.Unlock() 
 	if mock.SubFunc == nil { 
 		var ( 
 			out1 <-chan Queue 
 			out2 error 
 		) 
 		return out1, out2 
 	} 
 	return mock.SubFunc(topic) 
 }