Here is an update.

I have edited in Richard's text for __has_cpp_attribute (and added an #else
to his example, for improved realism/utility), and updated the entry for
[[deprecated]] to use it. I want to point out that I think this new feature
should be reviewed by EWG, at the very least, before we make this part of
the published SD-6.

I have also tweaked the entry for shared locking. There is an alternative
for N3659 that uses __has_include, but I have restored the (not very useful)
macro as it appears in the published SD-6. That's because (A) I feel that
the question of making a "breaking" change has not yet been discussed
adequately, and (B) the published recommendation is actually implementable,
even if it isn't ideal.

It seems to me that we're on a course that will not enable us to publish a
revised SD-6 before the Rapperswil meeting.

If anyone disagrees with my assessment, we could have a teleconference to
discuss things, probably sometime in May. Otherwise, perhaps we should just
meet in Rapperswil.

--
Clark Nelson            Vice chair, PL22.16 (ANSI C++ standard committee)
Intel Corporation       Chair, SG10 (C++ SG for feature-testing)
clark.nelson_at_[hidden]  Chair, CPLEX (C SG for parallel language extensions)