I think it comes down to a question of whether we want to guarantee that
these traits are treated as not being identifiers. In some sense, it's an
implementation detail that we model them as keywords -- and as it happens,
there are some circumstances in which we *don't* model them as keywords.
For
I've been using !__is_identifier to test for things like that. It seems to
be the most consistent way.
Is there some problem with this?
/Eric
On Thu, Jun 1, 2017 at 6:46 PM, Richard Smith wrote:
> On 31 May 2017 at 17:41, Eric Fiselier wrote:
>
>> I'm assuming libc++ should move to this trait
On 31 May 2017 at 17:41, Eric Fiselier wrote:
> I'm assuming libc++ should move to this trait instead?
>
Yes, that'd be a good idea. Though now that you mention it, I'm not sure we
have a good feature detection story for these builtins. Looks like a bunch
of the existing ones are (oddly) covered
I'm assuming libc++ should move to this trait instead?
/Eric
On Wed, May 31, 2017 at 6:28 PM, Richard Smith via cfe-commits <
cfe-commits@lists.llvm.org> wrote:
> Author: rsmith
> Date: Wed May 31 19:28:16 2017
> New Revision: 304376
>
> URL: http://llvm.org/viewvc/llvm-project?rev=304376&view=r
Author: rsmith
Date: Wed May 31 19:28:16 2017
New Revision: 304376
URL: http://llvm.org/viewvc/llvm-project?rev=304376&view=rev
Log:
PR33232: implement support for MSVC's __is_trivially_destructible trait.
Unlike the GCC-compatible __has_trivial_destructor trait, this one computes the
right answe
