Re: [PATCH v3 4/6] fsmonitor: add test cases for fsmonitor extension

[Ben Peart <peartben@gmail.com>]

On 5/31/2017 12:33 AM, Christian Couder wrote:
>>>> +test_expect_success 'refresh_index() invalidates fsmonitor cache' '
>>>> +       git config core.fsmonitor true &&
>>>> +       git config core.untrackedcache true &&
>>>> +       clean_repo &&
>>>> +       git status &&
>>>> +       test_path_is_missing marker &&
>>>> +       dirty_repo &&
>>>> +       write_script .git/hooks/query-fsmonitor<<-\EOF &&
>>>> +       :>marker
>>>> +       EOF
>>>> +       git add . &&
>>>> +       git commit -m "to reset" &&
>>>> +       git status &&
>>>> +       test_path_is_file marker &&
> 
> Ok so "marker" is there now.
> 
>>>> +       git reset HEAD~1 &&
>>>> +       git status >output &&
>>>> +       test_path_is_file marker &&
>>>
>>> You already checked that "marker" exists 3 lines above, and as far as
>>> I can see nothing could remove this file since the previous test, as
>>> the hook can only create it.
>>> So I wonder if something is missing or if this test is redundant.
>>
>> Testing it each time ensures it is being created when it is supposed to be
>> (ie when the test believes it is using the query-fsmonitor hook) and that it
>> isn't when it isn't supposed to be (ie when the hook should not be called).
> 
> I would agree with that if the "marker" file was removed after the
> previous "test_path_is_file marker", but I don't see any "clean_repo"
> or "rm marker" call that removes it.
> 

Got it.  I've added a call to "rm -f marker" to ensure the marker isn't 
left over from the previous status command.  I found another instance of 
where this could happen in "status doesn't detect unreported 
modifications" and fixed it as well.  Thanks!