`BOOST_TEST(p1 == p2)` doesn't always test `p1 == p2`

A user of Boost.Test has complained to me that the behavior of `BOOST_TEST(p1 == p2)` when `p1` and `p2` are pointers to `char` is surprising as it silently reinterprets the C++ expression `p1 == p2` to mean something else, namely, `strcmp(p1, p2) == 0`. This is, in this user's opinion, a significant defect that justifies not using Boost.Test. (I don't necessarily agree.)

I suspect that it's much too late now to change this behavior, as who knows how much existing code already relies on it, so I'm not sure what can be done to address this complaint.