pointer check in unit tests

[Stephan Bergmann]

On 04/11/2020 12:06, Mike Kaganski wrote:
> On 04.11.2020 13:51, Stephan Bergmann wrote:
>> That a reference is something completely different than a pointer 
>> shows e.g. in the rule that if a class has an implicit copy 
>> constructor and a non-static data member of reference type, then the 
>> copy constructor is defined as deleted (while there is no such rule if 
>> the data member is of pointer type).
> 
> I always had an impression that this is an obvious consequence of the 
> special initialization semantics of the references?

Sure, and maybe we are moving in circles.  My point in this thread has 
been that switching from T* to T& to rule out nullptr may be ill advice.