> and they left that many errors in. How many of those errors are visible to end users in a way that actually matters? A 'suspicious' floating point conversio

> and they left that many errors in. How many of those errors are visible to end users in a way that actually matters? A 'suspicious' floating point conversion in a display aspect ratio check? A memory leak in a program that's almost entirely interactively driven? I don't see anything else terribly huge on the list either. Software errors have an impact to the user and a cost to fix. These don't seem like errors where the ROI on the fix was likely to be positive, so maybe it's reasonable they were left in. You should also consider the errors that Microsoft _removed_ from the calculator before passing too much judgement on their software development process. After all, this is the same company that replaced IEEE754 floating point in calculator with an arbitrary precision library. (At considerably more expense than any of the bug fixes indicaetd by the attached article) https://blogs.msdn.microsoft.com/oldnewthing/20040525-00/?p=39193 https://blogs.msdn.microsoft.com/oldnewthing/20040525-00/?p=...