Skip to main content
7 events
when toggle format what by license comment
Oct 2, 2022 at 17:20 comment added Dave Cousineau I notice that the comments on that Michael Feathers article are basically arguing about this exact concept; MF says that a 1-dependency test is an integration test (it tests integration of your code with the dependency), whereas others are saying that they're still unit tests and are essential to tell if your systems actually work. MF's only argument against them seems to be the speed of running your tests.
Oct 2, 2022 at 16:51 comment added Dave Cousineau (It's possible that my code-base/code-style does not well support the kinds of testing that you're talking about, so my problems are specific to the way my code is written and you would design these systems differently in ways that I've not yet been able to.)
Oct 2, 2022 at 16:46 comment added Dave Cousineau My question is, what are "tests of 1-dependency", unit tests, or integration tests? Why aren't they differentiated from tests of 0-dependency. This seems to me like a crucial distinction, but in the communication of tests, this is never brought up. Always the distinction is unit vs integration, which seem to be 0-dependency and N-dependency respectively.
Oct 2, 2022 at 16:46 comment added Dave Cousineau I would say that the SpaceX example is (maybe) a test of 1-dependency, because you have a real engine with a mock rocket. (Not sure physical example is good because with software we have interfaces that let us pull some or all of the logic out of the "real" implementation. Theoretically we would mock the "heavy" parts (network, file access), and still be able to test the engine logic or rocket logic separately. It would be like knowing we have a working engine without using fuel.)
Oct 2, 2022 at 10:03 history edited Filip Milovanović CC BY-SA 4.0
added 125 characters in body
Oct 2, 2022 at 9:55 history edited Filip Milovanović CC BY-SA 4.0
added 36 characters in body
Oct 2, 2022 at 9:48 history answered Filip Milovanović CC BY-SA 4.0