> I think I get the intent with the report pattern. And I suspect it works best
> for arbitrary report types when the input (interface) is minimal and simple.
>
> Although I also think any report implementation class should be able to
> distinguish different events by string (or some other unique value).
I think any report implementation should implement handlers for each of the different events being reported on. There's a mechanism (pure virtual functions) in the language that enforces this at compile time. I prefer that to detecting a failure to match strings correctly in a test harness.
> I think I get the intent with the report pattern. And I suspect it works best
> for arbitrary report types when the input (interface) is minimal and simple.
>
> Although I also think any report implementation class should be able to
> distinguish different events by string (or some other unique value).
I think any report implementation should implement handlers for each of the different events being reported on. There's a mechanism (pure virtual functions) in the language that enforces this at compile time. I prefer that to detecting a failure to match strings correctly in a test harness.