Re: [Qemu-devel] [PATCH 4/5] checkpatch: bump most warnings to errors

[Thomas Huth]

On 10.08.2016 12:44, Paolo Bonzini wrote:
>>>>> diff --git a/scripts/checkpatch.pl b/scripts/checkpatch.pl
>>>>> index 714a000..ab08ca2 100755
>>>>> --- a/scripts/checkpatch.pl
>>>>> +++ b/scripts/checkpatch.pl
>>>>> @@ -1289,11 +1289,11 @@ sub process {
>>>>>                   # This is a signoff, if ugly, so do not double report.
>>>>>                   $signoff++;
>>>>>                   if (!($line =~ /^\s*Signed-off-by:/)) {
>>>>> -                         WARN("Signed-off-by: is the preferred form\n" .
>>>>> +                         ERROR("Signed-off-by: is the preferred form\n" .
>>>>>                                   $herecurr);
>>>>>                   }
>>>>
>>>> If you turn this into an ERROR, it's not the "preferred form" anymore,
>>>> but the "mandated form". So I'd suggest to either keep it as WARN or to
>>>> rephrase the message.
>>>
>>> What about:
>>>
>>>                                 ERROR("Signed-off-by: is spelled with
>>>                                 uppercase \"s\"\n" .
>>>                                         $herecurr);
>>
>> That would still be confusing if I'd spell it like "Signed-Off-BY", for
>> example.
> 
> 
> Right, so I guess "The correct form is \"Signed-off-by\"\n" is more precise.

Yes, that sounds better.

>> Maybe the outer check should simply not be case-insensitive, then you
>> could remove this check here completely?
> 
> The reason for that is to hide the "Missing Signed-off-by: line(s)" error.
> See here:
> 
>                         # This is a signoff, if ugly, so do not double report.
>                         $signoff++;
> ...
>         if ($is_patch && $chk_signoff && $signoff == 0) {
>                 ERROR("Missing Signed-off-by: line(s)\n");
>         }

Sure, but I think that error would be OK, too, since  most people should
be able to figure out that they spelled "signed-off-by" in a bad way
when they get a "Missing Signed-off-by: line" error.

 Thomas