Видео

Glad to hear that!

Also, your example actually seems to have centered around 1 real rule: Classes are proposed with size limits. I get that this centers around a specific business need, but there might be other things you come up with when talking about "what criteria are required to propose a class?" which directly impact the way you'd inevitably structure your entity. For example, what if we decided that all course proposal topics must be a subcategory of the business' core; in which case your new method would look like: ::propose(new Category($category), ClassSize::between(...)) perhaps. My point is, was it your intention to discuss the feature to this degree of isolation? Would you split this apart and refactor your tests after you add in the feature "Classes can only be a subcategory of business core"? Or do you try to account for all the rules that cannot be deferred in one feature? The reason why I find this important is because each example thereafter might reference rules categorically or not. For example, if you have both rules (class size AND subcategory) then some examples might not acknowledge certain rules through behaviour, because it isn't necessary for the grander scope of the example. For example, you could have your tests written exactly as they are ignoring the subcategory rule, and have another example: @Given core business categories are html, css and javascript @When a course is proposed with category 'lion taming' @Then the course is not considered viable. So this has some implications: class size tests which are viable possibly contradict this example: if the course category is not properly set but the class size is correct; then the course should still not be viable. So I suppose the moral of the story is to explicitly consider all of your constraints when developing your examples (maybe this is just a rubber duck, but I would still appreciate feedback).

Thanks mate. You saved my time.

There's no specific talk I'm afraid - the idea is that if we have: a) a narrow interface around an dependency that defines how we want to use it b) a fake implementation c) tests of that fake implementation Then we can re-use c) with our 'real' implementation to verify that the fake and real infrastructure behave the same way

@@ciaranmcnulty5713 thanks so much for replying. I think I get it. I'm not sure if I implemented the test the same as you would... The issue I had was I'm using a payment gateway. So it dictates my domain payment behaviour. I faked the dependency, but then I created an integration test with the real dependency. Not very elegantly, mind, I copied and pasted the fake integration test context into the real Integration context file, but just replaced all the services, including the database connection. Ah...I just got it. You wouldn't test the entire scenario, just the interaction with the service?

Yes just the interaction through the narrow interface (and the narrow interface is key, hide as much detail as you can). Grab me on the scotlandphp slack group if you want to chat

slack.scotlandphp.co.uk/