Builder.For should panic when called twice

[prometherion]

I know that a Controller should just control for a single Resource kind, this is an edgy-case.

With the following said

controller-runtime/pkg/builder/controller.go

Lines 65 to 68 in fdc6658

	// For defines the type of Object being *reconciled*, and configures the ControllerManagedBy to respond to create / delete /
	// update events by *reconciling the object*.
	// This is the equivalent of calling
	// Watches(&source.Kind{Type: apiType}, &handler.EnqueueRequestForObject{})

I'm considering For a shorthand for Watches that allows multiple inputs

controller-runtime/pkg/builder/controller.go

Lines 89 to 100 in fdc6658

	// Owns defines types of Objects being *generated* by the ControllerManagedBy, and configures the ControllerManagedBy to respond to
	// create / delete / update events by *reconciling the owner object*. This is the equivalent of calling
	// Watches(&source.Kind{Type: <ForType-forInput>}, &handler.EnqueueRequestForOwner{OwnerType: apiType, IsController: true})
	func (blder *Builder) Owns(object runtime.Object, opts ...OwnsOption) *Builder {
	input := OwnsInput{object: object}
	for _, opt := range opts {
	opt.ApplyToOwns(&input)
	}
	blder.ownsInput = append(blder.ownsInput, input)
	return blder
	}

but that's not true, because I got a single ForInput

controller-runtime/pkg/builder/controller.go

Lines 69 to 81 in fdc6658

	func (blder *Builder) For(object runtime.Object, opts ...ForOption) *Builder {
	if blder.forInput.object != nil {
	blder.forInput.err = fmt.Errorf("For(...) should only be called once, could not assign multiple objects for reconciliation")
	return blder
	}
	input := ForInput{object: object}
	for _, opt := range opts {
	opt.ApplyToFor(&input)
	}
	blder.forInput = input
	return blder
	}

Returning an error would be easier, although breaking the method chaining that is a de-facto standard with the Builder pattern: maybe is better to panic rather than making this non-predictable.

WDYT?

[dvob]

I just ran into the problem that the builder fails if Builder.For is not used. See the following example:
dvob@766c244

I think it would be the best if Builder.For would behave in the same way as Watches and Owns (so define forInput as slice []ForInput). This would fix the case when For is not used and would make it more predictable if it is used multiple times. Probably in most of the cases using Builder.For multiple times does not make sense but I don't think that a user should be stopped of doing that.

[dvob]

Just found #1173 and #1176 which adds an error if For is called multiple times.
But as said before: To me it would make more sense to allow multiple calls to For and not artificially restrict users.

[vincepri]

@alvaroaleman What do you think here? I'm ok having it panic given that's definitely a user error

[alvaroaleman]

#1176 already added an error that is returned on Build, that sounds good enough to me? (The error could probably be a bit more helpful and tell what was the pre-existing type and whats the new one, but thats nitpicking)

But as said before: To me it would make more sense to allow multiple calls to For and not artificially restrict users.

No. There is only one root type per controller. You have Watches to watch subordinate types.

[dvob]

Ok, but can we change the builder that it is possible to use it without For?
dvob@766c244

[fejta-bot]

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale

[fejta-bot]

Stale issues rot after 30d of inactivity.
Mark the issue as fresh with /remove-lifecycle rotten.
Rotten issues close after an additional 30d of inactivity.

If this issue is safe to close now please do so with /close.

Send feedback to sig-contributor-experience at kubernetes/community.
/lifecycle rotten

[fejta-bot]

Rotten issues close after 30d of inactivity.
Reopen the issue with /reopen.
Mark the issue as fresh with /remove-lifecycle rotten.

Send feedback to sig-contributor-experience at kubernetes/community.
/close

[k8s-ci-robot]

@fejta-bot: Closing this issue.

In response to this:

Rotten issues close after 30d of inactivity.
Reopen the issue with /reopen.
Mark the issue as fresh with /remove-lifecycle rotten.

Send feedback to sig-contributor-experience at kubernetes/community.
/close

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.