On Jan 26, 2011, at 8:51 AM, Sean Hunt wrote:
> On 11-01-26 10:39 AM, Douglas Gregor wrote:
>> Author: dgregor
>> Date: Wed Jan 26 09:39:56 2011
>> New Revision: 124293
>>
>> URL: http://llvm.org/viewvc/llvm-project?rev=124293&view=rev
>> Log:
>> Inline namespaces are always available in Clang. Rely on that without
>> testing via __has_feature, since __has_feature for C++0x features no
>> longer evaluates true in C++98/03 mode.
>>
>> Also, eliminate the redundant using directive. Inline namespaces make
>> their members visible in the enclosing namespace automatically.
>
> I think if we always provide inline namespaces, their __has_feature
> should be true irrespective of C++0x mode.
We always provide rvalue references, variadic templates, and a number of other
features, but complain about them with extension warnings. However, we
shouldn't advertise a C++0x feature to programs unless we're actually in C++0x
mode, since that entices programmers into writing non-conformant code.
- Doug
_______________________________________________
cfe-commits mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
