The chances that the author is not aware of what you are talking about is near 0, and what he is writing isn't an actual suggested specification it is musings a

The chances that the author is not aware of what you are talking about is near 0, and what he is writing isn't an actual suggested specification it is musings about pain points in the current way that things work. You are complaining about a single line of code that demonstrates a concept that cannot possibly be executed and there isn't any actual effort to make this line into something that can be executed. You are arguing that it is a bad idea for a reason that is completely orthogonal to the spirit of the post. So sure, implicit in that loop is the idea that iNaN would be treated subtly different than existing NaNs, but you are just responding to the text of an example that is written to be as simple and clear as possible to readers rather than the actual content of his message, which is nit picking at best.