Did you know that of the 2^64 possible floating-point numbers, 2^53 of
them are "nan" - and Python prints them all the same way?
While nans usually have all zeros in the sign bit and mantissa, this
isn't always true, and 'signaling' nans might trap or error. To help
distinguish such errors in e.g. CI logs, Hypothesis now prints "-nan"
for negative nans, and adds a comment like "# Saw 3 signaling NaNs" if
applicable.
The canonical version of these notes (with links) is on readthedocs.