Cool, that sounds pretty reasonable. Re. dynamic memory though, is there any chance you could support user-supplied malloc/free functions? Can't wait for the lib, best of luck getting it wrapped up and collecting the rest of that well-deserved bounty.