https://gcc.gnu.org/bugzilla/show_bug.cgi?id=81928
--- Comment #2 from Matthieu Brucher <matthieu.brucher at gmail dot com> --- No, I think the optimization is more than correct. The warning doesn't warn of the actual problem, if you check against nullptr, you won't even get a warning. Comparing this to anything from Null to nullptr should be an error, not be silently passed. Some people still think that it is valid because it worked in the past and because it's a bad anti pattern. So yes, it should be rejected because it's only point is to compare the pointer to nullptr, which is not sane. And it must be rejected because gcc in optimized mode considers this to be always not nullptr, and doesn't even warn that the expression is thus constant. The user doesn't know that he screwed big time.
