llvm dev - Jul 2016 - LibC++ failure on ARM

On 11 July 2016 at 15:30, Craig, Ben <ben.craig at codeaurora.org>
wrote:
> What C library is being used?  Maybe newlocale with the empty string
returns
> NULL with that C library.
(Ubuntu EGLIBC 2.19-0ubuntu6.3)

This is the exact same machine in the exact same configuration that we
were using for the old builder, just changed the master configuration.

> My suspicion is that the test is failing in the test block starting at line
> 48, where a std::locale is constructed with the empty string.
Why would it be breaking now and not before?

cheers,
--renato

On 7/11/2016 9:39 AM, Renato Golin wrote:
>> My suspicion is that the test is failing in the test block starting at
line
>> 48, where a std::locale is constructed with the empty string.
> Why would it be breaking now and not before?
No idea.  I work with the leads that I've got :)

Do you know if the old configuration claimed support for 
locale.en_US.UTF-8?  The failing test would have been skipped if that 
locale wasn't supported.
Are the old configurations and new configurations running similar 
quantities of tests?
>
> cheers,
> --renato
-- 
Employee of Qualcomm Innovation Center, Inc.
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a Linux
Foundation Collaborative Project

I changed the library order in response to a review comment [1], but can't
see how that's causing this failure.

@Renato: can you check on that box if reverting the library order actually
fixes the problem? Just to double check if that's indeed the cause.

Thanks.

/ Asiri

[1] http://reviews.llvm.org/D14293#inline-118681

On Mon, Jul 11, 2016 at 4:01 PM, Craig, Ben via llvm-dev <
llvm-dev at lists.llvm.org> wrote:
> On 7/11/2016 9:39 AM, Renato Golin wrote:
>
>> My suspicion is that the test is failing in the test block starting at
>>> line
>>> 48, where a std::locale is constructed with the empty string.
>>>
>> Why would it be breaking now and not before?
>>
> No idea.  I work with the leads that I've got :)
>
> Do you know if the old configuration claimed support for
> locale.en_US.UTF-8?  The failing test would have been skipped if that
> locale wasn't supported.
> Are the old configurations and new configurations running similar
> quantities of tests?
>
>
>> cheers,
>> --renato
>>
>
> --
> Employee of Qualcomm Innovation Center, Inc.
> Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a Linux
> Foundation Collaborative Project
>
> _______________________________________________
> LLVM Developers mailing list
> llvm-dev at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20160711/79614c11/attachment.html>