On 12/21/2015 09:28 AM, Andrei Alexandrescu wrote:




Two unnecessary pedantic corrections:

- One of the earlier slides has "Voldermort" with an extra 'r'. It should be Voldemort.

- At the end of one of the optimization presentations you said that the behaviour of unsigned integer overflow and *underflow* are well defined. Although what you meant is clear and that I've been using "underflow" in the same meaning until very recently, I now know that underflow is a condition only for floating point types.

Copying from Wikipedia, "Arithmetic underflow can occur when the true result of a floating point operation is smaller in magnitude (that is, closer to zero) than the smallest value representable as a normal floating point number in the target datatype."


Reply via email to