On 11/10/2019 09.54, Max Reitz wrote:
> On 10.10.19 20:42, Eric Blake wrote:
>> On 10/10/19 10:24 AM, Max Reitz wrote:
>>> Signed-off-by: Max Reitz <mreitz@redhat.com>
>>> ---
>>>   tests/qemu-iotests/common.filter | 8 ++++++--
>>>   1 file changed, 6 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/tests/qemu-iotests/common.filter
>>> b/tests/qemu-iotests/common.filter
>>> index 9f418b4881..cd42f5e7e3 100644
>>> --- a/tests/qemu-iotests/common.filter
>>> +++ b/tests/qemu-iotests/common.filter
>>> @@ -43,7 +43,8 @@ _filter_qom_path()
>>>   # replace occurrences of the actual TEST_DIR value with TEST_DIR
>>>   _filter_testdir()
>>>   {
>>> -    $SED -e "s#$TEST_DIR/#TEST_DIR/#g"
>>> +    $SED -e "s#$TEST_DIR/#TEST_DIR/#g" \
>>> +         -e "s#$SOCK_DIR/#SOCK_DIR/#g"
>>
>> Do we want to output a literal 'SOCK_DIR' (every test that uses it has
>> to update their expected output), or can we make this also output a
>> literal 'TEST_DIR' (output is a bit more confusing on which dir to look
>> in, but fewer files to touch)?  Your preference.
> 
> There’s another advantage to filtering it to be TEST_DIR, and that’s the
> fact that if $TEST_DIR and $SOCK_DIR are the same, we will always
> replace $SOCK_DIR by TEST_DIR.
> 
> But I still preferred filtering it to be SOCK_DIR, because that seemed
> to me like we would have done it had we had a SOCK_DIR from the start.

I also think that using SOCK_DIR is the better choice. It's a little bit
more churn now, but in the long run, it will help to avoid confusion,
and I think that's more important.

 Thomas