``core.tests.completion_test`` gives inconsistent results.

[brownnrl]

While running tests for re-factoring purposes, I've noticed that core.tests.completion_test gives inconsistent results over enough subsequent runs. I thought it was something I did, but it does so both on my branch and master. It seems to be just these particular two. I didn't want to slow myself down looking into it, thought maybe there was some known I/O dependency or history here where this was considered acceptable.

I think I've seen this both on the windows and linux boxes... these particular results were on a windows machine.

First Run:

>py.test core\tests\completion_test.py
============================= test session starts =============================
platform win32 -- Python 3.3.5 -- py-1.4.26 -- pytest-2.6.4
plugins: monkeyplus
collected 47 items

core/tests/completion_test.py .........F.....................................

================================== FAILURES ===================================
________________________ test_complete_latest_modified ________________________

app = <core.tests.base.TestApp object at 0x000000000422D358>

    @with_app(app_one_entry)
    def test_complete_latest_modified(app):
        # Always search the latest modified entries first for completion match.
        app.add_entry('31/10/2007', 'DepositFoo')
>       eq_(complete_etable(app, 'De', 'description'), 'positFoo')
E       AssertionError: 'posit' != 'positFoo'

core\tests\completion_test.py:147: AssertionError
=============== 1 failed, 46 passed, 1 warnings in 1.18 seconds ===============

Second Run:

> py.test core\tests\completion_test.py
============================= test session starts =============================
platform win32 -- Python 3.3.5 -- py-1.4.26 -- pytest-2.6.4
plugins: monkeyplus
collected 47 items

core/tests/completion_test.py ........FF.....................................

================================== FAILURES ===================================
_________________________ test_complete_goes_to_next __________________________

app = <core.tests.base.TestApp object at 0x0000000004279F28>

    @with_app(app_one_entry)
    def test_complete_goes_to_next(app):
        # As soon as a completion gets longer than the matched entry, find another one.
        app.add_entry('1/10/2007', description='Dep')
>       eq_(complete_etable(app, 'De', 'description'), 'p')
E       AssertionError: 'posit' != 'p'

core\tests\completion_test.py:140: AssertionError
________________________ test_complete_latest_modified ________________________

app = <core.tests.base.TestApp object at 0x0000000004291828>

    @with_app(app_one_entry)
    def test_complete_latest_modified(app):
        # Always search the latest modified entries first for completion match.
        app.add_entry('31/10/2007', 'DepositFoo')
>       eq_(complete_etable(app, 'De', 'description'), 'positFoo')
E       AssertionError: 'posit' != 'positFoo'

core\tests\completion_test.py:147: AssertionError
=============== 2 failed, 45 passed, 1 warnings in 1.14 seconds ===============

Third Run:

py.test core\tests\completion_test.py
============================= test session starts =============================
platform win32 -- Python 3.3.5 -- py-1.4.26 -- pytest-2.6.4
plugins: monkeyplus
collected 47 items

core/tests/completion_test.py ...............................................

==================== 47 passed, 1 warnings in 1.09 seconds ====================

[hsoft]

It's specific to windows. I think it has something to do with the precision of time.time() (completions are ordered by mtime).