This release changes how Hypothesis manages its search space in cases
where it generates redundant data. This should cause it to generate
significantly fewer duplicated examples (especially with short integer
ranges), and may cause it to produce more useful examples in some
cases (especially ones where there is a significant amount of
filtering).
The canonical version of these notes (with links) is on readthedocs.