Finally, we have gotten some concrete, specific feedback from someone who
wasn't involved in SG10; specifically, Matt Austern:

> Mostly looks fine to me. The one name here that I'm slightly concerned about
> is __cpp_lib_constexpr_functions, which is supposed to be defined for
> implementations that have additional constexpr support in <chrono>,
> <utility>, and containers. This is an awfully general name for something so
> specific; it seems to me there's a high risk of additional constexpr
> functions added in the future, where one would want to use that same name.
> Maybe the right fix is to remove that macro rather than try to find a better
> name, since it's rather a grab bag anyway. (Is there any reason to believe
> that those three sets of changes will be made simultaneously?)

I know that I once believed it to pretty important that the changes might be
applied to different headers at different times, and I don't specifically
remember being convinced otherwise. So I guess I missed this implication of
the change.

Does anybody want to defend the status quo?

Clark