Testing user interactions: the Spotter case study

Testing user interactions can be difficult, especially when the user interface relies on background processes.

An interesting case is Spotter. The interface resides in a dropdown and it relies on significant background processes: first for perfroming searches, and second for showing the selection preview.

To exemplify and test the interface, we use Scripter. Here is a scenario that is reasonably complicated:

we start a Spotter, and the first item is selected

we hover over the first item in the second category

then we hover over the item Ok action

and then we click on the button

spotterElementOnStartSampleActOnSecondCandidateWithDefaultObjectActOn
	<gtExample>
	| scripter actOnEvents actOn actOnHandler |
	scripter := self spotterElementOnStartSample.

	actOnEvents := OrderedCollection new.
	actOnHandler := GtSpotterByScripterExamplesActOnHandler new
			actOnEvents: actOnEvents.

	scripter element spotter announcer weak
		when: GtSpotterObjectActOn
		send: #onActed:
		to: actOnHandler.

	scripter
		mouseMoveOverStep: [ :s | 
			s
				id: GtSpotterCandidateElementId index: 2 ].

	scripter
		mouseMoveOverStep: [ :s | 
			s
				id: GtSpotterCandidateElementId index: 2;
				id: GtSpotterCandidateActOnButtonId ].

	scripter
		clickStep: [ :s | 
			s
				id: GtSpotterCandidateElementId index: 2;
				id: GtSpotterCandidateActOnButtonId ].

	self assert: actOnEvents size equals: 1.

	actOn := actOnEvents first.

	self assert: actOn isActedUpon equals: true.
	self assert: actOn spotterSearch title equals: 'Priority 20'.
	self assert: actOn spotterStep equals: scripter element spotter steps first.
	self assert: actOn rawObject equals: 20.
	self assert: actOn objectToSend equals: 20.
	self assert: actOn toolToSend class equals: GtInspectorTool.
	self assert: actOn toolToSend object equals: 20.

	^ scripter

Scripter not only allows us to express all this logic and to deal with the background behavior transparently, but it also provides a preview of the actions.

This allows us to go back and forth between the script and the preview to guide the actual testing scenario.