On 14.09.20 12:50, Thomas Huth wrote:
> On 14/09/2020 11.19, Max Reitz wrote:
>> On 12.09.20 14:14, Thomas Huth wrote:
>>> macOS is shipped with a very old version of the bash (3.2), which
>>> is currently not suitable for running the iotests anymore. Add
>>> a check to skip the iotests in this case - if someone still wants
>>> to run the iotests on macOS, they can install a newer version from
>>> homebrew, for example.
>>>
>>> Signed-off-by: Thomas Huth <thuth@redhat.com>
>>> ---
>>>  tests/check-block.sh | 5 +++++
>>>  1 file changed, 5 insertions(+)
>>>
>>> diff --git a/tests/check-block.sh b/tests/check-block.sh
>>> index 8e29c868e5..bfe1630c1e 100755
>>> --- a/tests/check-block.sh
>>> +++ b/tests/check-block.sh
>>> @@ -46,6 +46,11 @@ if ! command -v bash >/dev/null 2>&1 ; then
>>>      exit 0
>>>  fi
>>>  
>>> +if bash --version | grep 'GNU bash, version [123]' > /dev/null 2>&1 ; then
>>
>> grep -q instead of the redirections, perhaps?
>>
>> But more importantly, I think this needs a LANG=C prefix.  (If I expand
>> the rejected major versions to [12345], it doesn’t skip without a
>> prefix, because the string reads “GNU bash, Version 5...” here in
>> LANG=de_DE.UTF-8.)
> 
> Ouch, ok. But actually, I'm not quite sure anymore whether the patch is
> really required. I ran into the "readlink -f" problem and thought that
> it occurred due to the ancient version of bash on macOS, but as a I now
> know, readlink is a separate program and not a bash built-in, so it's a
> different issue... thus let's skip this patch here for now until we hit
> a real issue with bash again.

Yes, I had hoped this patch would fix that issue.  Or perhaps at least
hide it, because if you have a newer bash, chances are your readlink has
-f, too.

So should we just effectively revert b1cbc33a397 if readlink -f didn’t
work, i.e. check "$?" and on failure use $PWD as it was before b1cbc33a397?

Max