Take a look at the stream of commits, they should make my point clearer. The main problem is the lack of feedback when I make changes to add/modify a behavior. In the GOOS book the use of a mocking framework is paired with E2E tests, which I rarely see around because they're costly. Furthermore, mocking frameworks are a sharp knife and I've seen some horrible usages around.
Q: But without a mocking framework you can't do outside-in TDD!
A: Nope, not related. You can still do outside-in using manual fakes, I do!
Q: The test using manual fakes is more opaque
A: I respectfully disagree. Tests should not reveal implementation details but express an observable behavior, that's what makes refactoring sustainable
Q: The tests that reuse a manual fakes are a little more coupled, meaning that if you make a mistake in the "InMemory..." fake implementation a bunch of tests will break
A: Yes, that's true and it's a trade-off I'm willing to accept because it hasn't been a problem so far...
Do you have a question or observation to make? Post an issue and I'll try to answer