> The patch you have now looks good but I think a more universal solution might 
> be better since these tests will fail whenever -f-short-enums is used. 
> Something like:
I'm not sure I understand... This works on ARM for both -fshort-enums and 
-fno-short-enums.

>     test_make_signed<const Enum, std::conditional<sizeof(Enum) == 
> sizeof(int), const int, const signed char>::type>()

That seems better on the surface, but assumes that short enums are char sized, 
which might not be the case on all platforms.  I think that the version that 
checks __ARM_SIZEOF_MINIMAL_ENUM does a better job of calling out the fact that 
it's a weird case.

> Out of curiosity how are you getting the -fshort-enums flag into lit to test?
I need ABI compatibility with a bare-metal GCC toolchain and that toolchain 
defaults bare-metal AAPCS targets to -fshort-enums, so my driver sets it.

http://reviews.llvm.org/D4869



_______________________________________________
cfe-commits mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits

Reply via email to