Cannot unpack non-iterable MultipleResults object

[vxgmichel]

Hi !

I ran into a problem with the falsifying examples produced by stateful testing with multiple results.

Consider the following state machine:

from hypothesis.stateful import Bundle, RuleBasedStateMachine
from hypothesis.stateful import rule, multiple, initialize


class DatabaseComparison(RuleBasedStateMachine):

    values = Bundle("values")

    @initialize(target=values)
    def produce(self):
        return multiple(1, 2, 3)

    @rule(v=values)
    def check(self, v):
        assert v == 1


TestDBComparison = DatabaseComparison.TestCase

It legitimately fails with the following info:

E       AssertionError: assert 3 == 1

test_hypo.py:16: AssertionError
---------------------- Hypothesis --------------------
Falsifying example:
state = DatabaseComparison()
v1, v2, v3 = state.produce()
state.check(v=v3)
state.teardown()

Let's put the falsifying example into a dedicated test:

def test_reproduce():
    state = DatabaseComparison()
    v1, v2, v3 = state.produce()
    state.check(v=v3)
    state.teardown()

The problem is that the test fails with the following error:

    def test_reproduce():
        state = DatabaseComparison()
>       v1, v2, v3 = state.produce()
E       TypeError: cannot unpack non-iterable MultipleResults object

test_hypo.py:24: TypeError

This can be fixed with the following change:

def test_reproduce_fixed():
    state = DatabaseComparison()
    v1, v2, v3 = state.produce().values
    state.check(v=v3)
    state.teardown()

Which does produce the right error:

E       assert 3 == 1

test_hypo.py:16: AssertionError

[Zac-HD]

Definitely a bug - fortunately it's easy to fix, just adding an __iter__ method to

hypothesis/hypothesis-python/src/hypothesis/stateful.py

Lines 355 to 357 in 47f82a1

	@attr.s()
	class MultipleResults:
	values = attr.ib()

pasting your test case above into hypothesis-python/tests/cover/test_stateful.py.

Would you like to open a PR, so python-trio/hypothesis-trio#5 doesn't need a workaround?

[vxgmichel]

@Zac-HD

Would you like to open a PR, so python-trio/hypothesis-trio#5 doesn't need a workaround?

Done: #2312 :)