Add refresh button to ResourceLocator field-type

[niklasnatter]

Q	A
Bug fix?	no
New feature?	yes
BC breaks?	no
Deprecations?	no
Fixed tickets	fixes #4221, fixes #5547
Related issues/PRs	#5340
License	MIT

What's in this PR?

This PR adds a Refresh URL button to the ResourceLocator field-type. The button allows to regenerate the URL based on the current values of the Form.

Why?

Because it is frequently requested feature 🙂
See #4221 and #5340.

[danrot]

I have just tested it, and to me it feels a little bit weird from a user perspective...

Let's assume we have just the title and the URL field. The URL field uses the value from the title field to generate the URL. I've realized two weird behaviors:

1. The refresh button won't be activated until I the user blurs the title field.
2. The refresh button also activates, if the user focuses the title field and blurs it without applying any change.

The first one requires an additional click, which should not be necessary IMO, and the second one might be confusing to the user, because nothing is happening after the button click.

That made me wonder: Wouldn't it make sense to set the enableRefreshButton when a change is happening? We could do that by adding a addChangeFieldHandler. I guess alternatively that could also be done by using a @computed value for the url parts and use an autorun.

[niklasnatter]

Yes, I think the behaviour is not perfect too. I think adding a addChangeFieldHandler would be nice, but I am not sure how to implement it.

As far as I see, the dataPath of the field is passed to the given FinishFieldHandler. Also, the the FormStoreInterface::finishField method has a dataPath parameter. In contrast, the FormStoreInterface::change method uses a name parameter. Is there a specific reason for this disparity? Or would we implement the FormStoreInterface::change method with a dataPath too, if we would implement it again? WhatShould the ChangeFieldHandler receive a name or a dataPath? 🤔

Furthermore, the Form container calls the FormInspector::finishField method. The FormInspector then calls the FormStoreInterface::finishField method. Should we implement a similar FormInspector::changeField method? That feels kinda wrong to me as "Inspector" sounds like it is a read-only thing, doesnt it?

One solution for the second problem would be adding a FormStoreInterface::notifyFieldFinish and a FormStoreInterface::notifyFieldChange method and deprecating the FormStoreInterface::finishField method. But maybe I am just overthinking this? 😅

[niklasnatter]

This PR depends on #5558, therefore I rebased it temporarily. Please remind me to rebase this back onto master before merging it in case I forget about it

[niklasnatter]

I do not understand why circleci is failing, I have not adjusted this files as far as I see 😕

[alexander-schranz]

A error happens on the homepage when the partsChangeDisposer is not set.

[danrot]

I do not understand why circleci is failing, I have not adjusted this files as far as I see

Probably related to #5569, but there is also a jest fix on its way. Not sure if we should merge #5569 or wait for the fix.

[alexander-schranz]

Works now as expected. Thx for adding the unmount test 👍