[cmake-developers] C++11 and target_compiler_feature proposal

Rolf Eike Beer eike at sf-mail.de
Fri Oct 11 14:27:02 EDT 2013


Am Freitag, 11. Oktober 2013, 14:14:34 schrieb Brad King:
> On 10/11/2013 02:04 PM, Sean McBride wrote:
> > On Fri, 11 Oct 2013 13:56:23 -0400, Brad King said:
> >> No, sorry.  What does "combined gcc+clang" mean?
> > 
> > He's probably referring to so-called 'llvm-gcc':
> Ah, thanks Sean.  The machines are at least:
> 
> * dashmacmini4 has Xcode 3.2 with llvm-gcc-4.2:
> 
>    #define __GNUC__ 4
>    #define __GNUC_MINOR__ 2
>    #define __GNUC_PATCHLEVEL__ 1
>    #define __APPLE_CC__ 5646
>    #define __llvm__ 1
> 
> * dashmacmini5 has Xcode 4.1 with llvm-gcc-4.2:
> 
>    #define __GNUC__ 4
>    #define __GNUC_MINOR__ 2
>    #define __GNUC_PATCHLEVEL__ 1
>    #define __APPLE_CC__ 5658
>    #define __llvm__ 1
> 
> They do not define __clang__, and CMake's compiler id detection
> does not pay attention to __llvm__.  I do not know what feature
> sets they support, but it is probably safe to assume GNU 4.2.1
> feature support at least.

Yes, that works, but only because Clang "by accident" supports more features 
than gcc.

But I think this is a good example of a system where the user could use the 
"force test results to come out of compilation" to get more meaningful 
results.

Eike
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part.
URL: <http://public.kitware.com/pipermail/cmake-developers/attachments/20131011/ef95096f/attachment.sig>


More information about the cmake-developers mailing list