An example that I think might show an opportunity to enhance Python Flexmock.

signature_change.py

from flexmock import flexmock

# foo() is the code under test
# other.bar() is a dependency we are mocking
def foo(other):
    other.bar(1, 2)


class OldVersionOfDep():
    def bar(self, x, y):
        print x, y


def test_foo_with_old():
    other = OldVersionOfDep()
    (flexmock(other)
        .should_receive('bar')
        .with_args(1, 2)
        .replace_with(lambda x, y: None))
    foo(other)


class NewVersionOfDep():
    def bar(self, x, y, z):
        print x, y, z


def test_foo_with_new():
    other = NewVersionOfDep()
    (flexmock(other)
        .should_receive('bar')
        .with_args(1, 2)
        .replace_with(lambda x, y: None))
    foo(other)


print "Old version of dependency."
test_foo_with_old()  # Passes.
foo(OldVersionOfDep())  # Works.
# -> 1 2
print "New version of dependency."
test_foo_with_new()  # Passes, but I wish it didn't...
foo(NewVersionOfDep())  # Broken! (Flexmock could catch this!)
# -> TypeError: bar() takes exactly 4 arguments (3 given)

The point here is that when the things you mock change there signatures, it would be nice to get an error. I think that's a common class of bug and it would be reasonable cheap to detect.

Feature implemented! Go Python Flexmock!

has207/flexmock#79 (comment)

yellowbus:fmock luke$ python test.py
Old version of dependency.
1 2
New version of dependency.
Traceback (most recent call last):
  File "test.py", line 42, in <module>
    test_foo_with_new()  # Passes, but I wish it didn't...
  File "test.py", line 32, in test_foo_with_new
    .with_args(1, 2)
  File "/Users/luke/src/fmock/src/flexmock/flexmock.py", line 349, in with_args
    self._verify_signature_match(*kargs, **kwargs)
  File "/Users/luke/src/fmock/src/flexmock/flexmock.py", line 250, in _verify_signature_match
    (self.name, minimum, total_positional))
flexmock.FlexmockError: bar requires at least 3 arguments, expectation provided 2