copying falsifying example into @example() fails to reproduce test failure due to pprinter sorting dict keys

[andreareina]

Some code depends on the order of dict iteration so sorting the keys before printing makes reproducing a failing test more difficult.

Related to #2913 ?

Sample code:

import hypothesis as h
import hypothesis.strategies as st

@h.given(st.dictionaries(st.integers(), st.integers()))
def test(d):
    h.note(f"{d=}")
    assert sorted(d) == list(d)

Output: falsifying example report gives an input that if copied verbatim, would actually pass the test

=========================================================================================== test session starts ===========================================================================================
platform linux -- Python 3.8.11, pytest-5.4.3, py-1.11.0, pluggy-0.13.1
rootdir: [snip]/hypothesis-dictionary-order
plugins: hypothesis-6.47.0
collected 1 item                                                                                                                                                                                          

test.py F                                                                                                                                                                                           [100%]

================================================================================================ FAILURES =================================================================================================
__________________________________________________________________________________________________ test ___________________________________________________________________________________________________

    @h.given(st.dictionaries(st.integers(), st.integers()))
>   def test(d):

test.py:5: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

d = {0: 0, 1: 0}

    @h.given(st.dictionaries(st.integers(), st.integers()))
    def test(d):
        h.note(f"{d=}")
>       assert sorted(d) == list(d)
E       assert [0, 1] == [1, 0]
E         At index 0 diff: 0 != 1
E         Use -v to get the full diff

test.py:7: AssertionError
----------------------------------------------------------------------------------------------- Hypothesis ------------------------------------------------------------------------------------------------
Falsifying example: test(
    d={0: 0, 1: 0},
)
d={1: 0, 0: 0}
========================================================================================= short test summary info =========================================================================================
FAILED test.py::test - assert [0, 1] == [1, 0]

Expected output: falsifying example should reflect the dict key order as actually passed into the test function

[Zac-HD]

Ooh, thanks for the report - this was OK to do on old versions of Python, but not now that dict iteration order is stable.

[richardscholtens]

This PR was kindly supported by Ordina Pythoneers.