[llvm-commits] Is AlignOfTest.cpp intentionally testing ambiguous inheritance?

[Sean Silva]

Chandler has already implicitly approved the direction and this shifts
some load off his back, so go ahead. I'm not knowledgeable about these
preprocessor hacks so I can't really give any review w.r.t. that.

-- Sean Silva

On Fri, Nov 16, 2012 at 11:30 PM, Joe Abbey <joe.abbey at gmail.com> wrote:
>
> On Nov 16, 2012, at 8:13 PM, Sean Silva <silvas at purdue.edu> wrote:
>
>>> I'm happy to add pragmas to suppress the warnings in this file, we
>>> have them for some compilers already. I'd rather nat remove the test
>>> just because I don't know how else to get coverage here besides
>>> somewhat exhaustive testing of various patterns.
>>
>> I think this would be appreciated just to keep the build a little bit
>> cleaner. Thanks!
>>
>> -- Sean Silva
>
>
> Suppressing all warnings from GCC in AlignOfTest.cpp
>
> Index: ../unittests/Support/AlignOfTest.cpp
> ===================================================================
> --- ../unittests/Support/AlignOfTest.cpp        (revision 168247)
> +++ ../unittests/Support/AlignOfTest.cpp        (working copy)
> @@ -22,6 +22,18 @@
>  #pragma warning(disable:4584)
>  #endif
>
> +// Suppress direct base ‘{anonymous}::S1’ inaccessible in ‘{anonymous}::D9’
> +// due to ambiguity warning.
> +//
> +// Pragma based warning suppression was introduced in GGC 4.2.  Additionally
> +// this warning is "enabled by default".  The warning still appears if -Wall is
> +// suppressed.  Apparently GCC suppresses it when -w is specifed, which is odd.
> +// At any rate, clang on the other hand gripes about -Wunknown-pragma, so
> +// leaving clang out of this.
> +#if ((__GNUC__ * 100) + __GNUC_MINOR__) >= 402 && !defined(__clang__)
> +#pragma GCC diagnostic warning "-w"
> +#endif
> +
>  // Define some fixed alignment types to use in these tests.
>  #if __has_feature(cxx_alignas)
>  struct alignas(1) A1 { };
>
> Ok to commit?
>
> Joe