In discussing the benefits of a skinny controller on p. 256, the authors say “That allows us to unit test it in isolation later if we want”. Do they mean in the context of the book, or in general? Because models should be 100% unit tested, in my opinion.
You’re right, and we agree. We’ve tried to keep the book focussed on using Cucumber for acceptance testing, rather than the whole outside-in flow. We’ve tried to signpost readers to books like GOOS and The RSpec Book if they’re interested in this.