Cover example differs

[isidroas]

When doing the example of https://crosshair.readthedocs.io/en/latest/cover.html , I noticed that the result is different:

$ crosshair cover foo.py
average([0.0], default=float("inf"))

To get the same results I had to increase the iterations:

$ crosshair cover --max_uninteresting_iterations 15 foo.py
average([0.0], default=float("inf"))
average([], default=0.0)

$ crosshair cover --max_uninteresting_iterations 19 foo.py
average([0.0], default=float("inf"))
average([], default=0.0)
average([], default=None)

offtopic: Thank you for this very curious library!

[pschanely]

Ah, yes, as of recently, CrossHair attempts true floating point semantics. That's good, but it's made this example take longer. I'll update the example soon. (or maybe tweak the heuristics)

offtopic: Thank you for this very curious library!

😊 If you end up trying it out, I would love to hear more about what works well for you, and what doesn't!

[pschanely]

Update: I am delaying a documentation fix for this slightly, under the hope that one of my unrelated arcs of work makes it not necessary. 😄