r189532 - Mode is now handled as a non-inheritable attribute, and CUDADevice & CUDAHost are now handled as inheritable attributes. In all three cases, this makes the processing behavior more consistent with the declared behavior in Attr.td.

[Aaron Ballman]

On Wed, Aug 28, 2013 at 8:19 PM, Richard Smith <richard at metafoo.co.uk> wrote:
> Digging into this further, I think this is actually a non-functional change.
> There's no real difference between ProcessNonInheritableDeclAttr and
> ProcessInheritableDeclAttr.

There is because combining the two cause a lot of test case failures.
But I would be *delighted* if we could get rid of the two because it
would make consolidated error checking much, much easier.

~Aaron

>
>
> On Wed, Aug 28, 2013 at 5:11 PM, Eli Friedman <eli.friedman at gmail.com>
> wrote:
>>
>> On Wed, Aug 28, 2013 at 4:23 PM, Aaron Ballman <aaron at aaronballman.com>
>> wrote:
>>>
>>> Would this make a suitable test case?
>>>
>>> typedef int i16_1 __attribute((mode(HI)));
>>> typedef i16_1  i16_redecl;
>>
>>
>> That isn't a redeclaration. :)
>>
>> Anyway, I'm not sure you actually can write a meaningful testcase; the
>> ModeAttr AST node doesn't actually have any semantics.
>>
>> -Eli
>
>