disabling auto-verifcation

[joprice]

I commented here #671 (comment) on a situation where tests that return a result type show the expectation unmet error instead of the actual test failure. The suggestion is to use panics, but I would have to rewrite hundreds of tests that currently essentially use the ? operator as a form of assertion. In order to preserve this structure, perhaps they could be wrapped in a macro that runs the test function and unwraps the result, with the mocks being created outside that helper. I think that would have the same behavior, but unsure. I thought perhaps there could be a feature like "manual_verification", where the drop impl is a no-op and you need to call a verify function yourself, like mock in other languages lke java's mockito.

[asomers]

There is a manual verification function; it's called checkpoint. But that doesn't eliminate checking call counts on drop; it augments it. In your proposal, when would the "manual_verification" occur? It can only occur before the mock object is dropped. And in the example you gave in issue 671, the object got dropped before test test otherwise paniced, so you would still have the same problem. Could you provide an example of the suggested usage of this "manual_verification" feature?