In the Utility functions section:

fsub():
Exact addition can be inefficient and may be impossible to perform with large magnitude differences:

It should be:

Exact subtraction can be inefficient and may be impossible to perform with large magnitude differences:

mag():
Quick logarithmic magnitude estimate of a number…

But the implementation should return the exponent exactly (if I am correct), maybe change to:

Returns an integer or infinity m such that 2^m-1 ≤ |x| < 2^m.