Yup, right on. We found a bug that crops up when a certain compiler (proprietary, not MSVC) tries to reuse the result of a float comparison originally made in a ternary expression.
The compiler ends up omitting most of the rest of the function's assembly, which breaks things as much as you'd expect. (For bonus fun, our ternary expression was deep under about four layers of macro.)
The compiler ends up omitting most of the rest of the function's assembly, which breaks things as much as you'd expect. (For bonus fun, our ternary expression was deep under about four layers of macro.)