Re: LIC_FILES_CHKSUM fatal?

[Gary Thomas <gary@mlbassoc.com>]

On 05/16/2011 01:57 PM, Richard Purdie wrote:
> On Mon, 2011-05-16 at 17:24 +0100, Phil Blundell wrote:
>> This change, from commit d2d5456cd3b3bd3e52a5dedccca4d46e3a7986d1:
>>
>> diff --git a/meta/classes/insane.bbclass b/meta/classes/insane.bbclass
>> index 225530a..71ed5b6 100644
>> --- a/meta/classes/insane.bbclass
>> +++ b/meta/classes/insane.bbclass
>> @@ -349,7 +349,7 @@ def package_qa_check_license(workdir, d):
>>           # just throw a warning now. Once licensing data in entered for enough of the recipes,
>>           # this will be converted into error and False will be returned.
>>           bb.error(pn + ": Recipe file does not have license file information (LIC_FILES_CHKSUM)")
>> -        return True
>> +        return False
>>
>> means that, contrary to what the comment says, lack of LIC_FILES_CHKSUM
>> is now a fatal error.  Was this a deliberate change or an oversight?
>>
>
> It was a deliberate change, the comment was missed in the change though.
>
> Yes, its annoying and we can have some way to turn it off but for at
> least the core, having them present is a good thing.

Even if this check doesn't treat this as fatal, my experience has
been that do_configure() fails if LIC_FILES_CHKSUM is missing.

-- 
------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------