There are instances where a 32 bit macro is used to assign a value to a 16 bit variable. In those cases this is done without any explicit narrowing on one hand, and it generates bit true differences (on other hand) for systems without 8 bit addressing and who doesn't wrap at 16 bits.