Eric Niebler

Unrelated: I wonder if the stdlib would be improved by the addition of alias templates for function and array types to hide C's awful declarator syntax: ```c++ template using c_function...

Devil's advocate here: it would be surprising for some code if a `pair` didn't store a `T` and a `U`.

> `get` won't work without an additional overload (which [namespace.std] bans). It would need a little work. In the working draft, the index-based overloads are specified to use `tuple_element` to...

Dropping to P2.

I would love for the need for `__commom_tuple` to die. I would also like the `tuple` and `pair` assignment operators to accept any type that is compatible with the structured...

[Compiler-explorer](https://github.com/mattgodbolt/compiler-explorer) supports libraries. I wager there's some code and infra that could be shared.

I don't recall if the signed-ness of `view_interface::size` was discussed in LEWG. @CaseyCarter, do you recall? I'm inclined to believe this is a bug in the standard. I'd rather not...

This is actually how the constraints of `for_each` are currently specified in the C++20 draft. We'll need to do some hard thinking about whether we should loosen the constraints of...

I don't think this is a bug. The elements must be partially ordered, IIRC. The presence of inf in the range violates the algorithm's preconditions.

For `a` and `b` to be partially ordered relative to each other, at most one of these must be true: `a < b` `b < a` That's not true for...