On Mon, Nov 16, 2015 at 1:12 PM, Aaron Ballman <aaron_at_[hidden]>
wrote:

> On Fri, Nov 13, 2015 at 8:58 PM, Nelson, Clark <clark.nelson_at_[hidden]>
> wrote:
> > Here is a draft of SD-6, updated with decisions from the Kona meeting.
> But
> > naturally, there are lot of places I mark guesses that I have made.
> >
> > Unfortunately, because the redlining is relative to the published SD-6,
> it
> > might be less than obvious what is really new -- except that we changed
> from
> > N-numbers to P-numbers at this meeting.
> >
> > As always, corrections and contributions are most earnestly welcomed.
>
> I am slightly confused about:
>
> N4266: Attributes for namespaces and enumerators
>
> Example:
>
> enum {
> old_val
> #if __cpp_enumerator_attributes
> [[deprecated]]
> #endif
> , new_val };
>
> The notion of feature testing macros for C++ attributes was not voted
> favorably by EWG in Kona, and what I understand from reading the
> discussion (which could be me horribly misunderstanding the feeling in
> the room) was that there was insufficient motivation for such a thing.
> However, this is a feature testing macro... for an attribute, just
> with a different spelling than what SG-10 proposed.
>
> If we need this feature testing macro (which I believe we do), can we
> use this as a motivating case for EWG to reconsider as to why
> __has_cpp_attribute is valuable?


This is feature-testing for the ability to apply an attribute to an
enumeration, not for the [[deprecated]] attribute itself. Without the
#ifdef, the above code would fail to parse in compilers that do not
implement the change allowing attributes on enumerators.