-Wdouble-promotion
This warning informs of float-to-double conversions. The best solution seems to be to do math *either* in 32-bit *or* in 64-bit, and only to cross over when absolutely necessary.
-Wnon-virtual-dtor