What is better practice in Cucumber-jvm

Question

We are at odds in the office over what is the better way to structure our cucumber test suite. This is in the context of clicking radio buttons in forms. The two approaches are:

1. Use the page object model. Have the radio buttons defined as web elements in the page object and pass in the value to select the option as input in the feature file.
2. Pass a selector (the element attribute "name") and value as inputs in the feature file. By passing the selector and value, all we would need is a method that clicks on the option dynamically according to the selector and value passed.

Option 1:

Pros:

• Since the webElements are defined in the PageObject, you don't have to pass the selector each usage. You only have to use the variable.
• Better for the long run because you only have to change/remove the WebElement in the PageObject rather than at every usage
• Using a variable allows you to utilize the IDEs built in capabilities. Refactoring, removing, find usages of, etc.
• More inline with the Page Object Model which we have outlined thus being (in my opinion) better practice

Option 2:

Pros:

• Don't have to define every single element thus reducing bloat in code (Which my boss says isn't a priority)
• Development is smoother because you don't have to be concerned about defining each new member variable, you just have to pass the selector as input.

In my opinion, option 2 might be better for the short term but as we develop more and scale up, having webElements defined in classes will be more beneficial. And I'm also thinking that development using option 2 is actually going to be clunkier since you need to find the selector each time you use it.

We both think that the other is objectively worse but can't come to an agreement. Is either one objectively better? Which is better practice from an industry standard point of view?

Answer 1

I'd choose page objects instead of selectors in the feature files without a second thought because feature files and page object should represent different levels of abstraction. Page objects can of course share common functionality any way you see fit to reduce bloat.