[PATCH] [ARM] Fix AArch32 and pre-v8 poly types to be unsigned

[Bob Wilson]

On Oct 11, 2013, at 9:07 AM, Bradley Smith <bradley.smith at arm.com> wrote:

> Hi Bob,
> 
> Since I believe it was you who objected somewhat to this last time around,
> do you have any opinion on this? Thanks.
> 
> Regards,
> Bradley Smith

So it hasn’t even been proposed for GCC yet?  Why not?

It seems unlikely but if there’s a chance that the GCC community may object to an incompatible ABI break like this, then wouldn’t it be better to wait?

> 
>>> On 2 October 2013 16:09, Bradley Smith <bradley.smith at arm.com> wrote:
>>>> All poly types should be of unsigned type,
>>> 
>>> I proposed this back in January, and I think we reached the
>> conclusion
>>> that this change should be coordinated with GCC since it's
>> potentially
>>> ABI-breaking.
>>> 
>>> Do you know if there's been any progress on the GCC side? I think
>> they
>>> (Richard Earnshaw & Tejas) were intending to wait until 4.8.0 had
>> been
>>> released before proposing the change there.
>> 
>> Unfortunately not much it seems, it's still very much on their TODO
>> list
>> for 4.8 so will happen at some point, we're just not sure exactly when.
>> Given this is now specified in the ACLE as unsigned so has to change in
>> GCC,
>> it might be nice for clang to take the lead on this one rather than
>> waiting
>> for GCC.
>> 
>> Also as you stated in your previous thread, which I just found, it's
>> extraordinarily unlikely that many people if anybody will actually be
>> affected
>> by this change differing across clang and GCC until GCC changes it.
> 
>