Re: [PATCH] Correctly use the bitbake variable S from now on

[Jan Kiszka <jan.kiszka@siemens.com>]

On 2018-04-19 12:36, Henning Schild wrote:
>>> diff --git a/meta/classes/patch.bbclass b/meta/classes/patch.bbclass
>>> index 26a0c81..1559359 100644
>>> --- a/meta/classes/patch.bbclass
>>> +++ b/meta/classes/patch.bbclass
>>> @@ -5,7 +5,7 @@ python do_patch() {
>>>      import subprocess
>>>  
>>>      workdir = d.getVar("WORKDIR", True) + "/"
>>> -    src_dir = workdir + (d.getVar("S", True) or "")
>>> +    src_dir = (d.getVar("S", True) or "")  
>>
>> That's not equivalent, but I'm undecided if it matters. S should never
>> be unset (bitbake.conf holds the default). So we could just pull S and
>> let the build fail if that assumption should ever be wrong.
> 
> I was wondering why the or "" was there in the first place, S was
> always set even before recipes decided to overwrite it. Some python
> sniplets seem to return Null on some intermediate steps when using
> layers. Might that be why you introduced the 'or ""'?

Problem is that I don't remember doing this intentionally. But if that
were the case (intermediate Null state), it should not longer harm us
now (no more "x + Null").

Jan

-- 
Siemens AG, Corporate Technology, CT RDA IOT SES-DE
Corporate Competence Center Embedded Linux