> I have P1030R3 on the list of papers to schedule for Belfast [1], but I
> haven't finished the schedule yet. We're going to be short on time, but
> I plan to make time for this paper. Some things that you could do to help:
>
> 1) Have a brief introduction prepared that goes over the changes since
> R2 and the affect those changes have had on users (e.g., some details
> like those you describe below).

Easy to review the changes. Reviewing impact would be much harder. The
reason is that the reference implementation only caught up to the paper
two weeks ago, and insufficient time has elapsed to see on what users
complain.

> 2) Review the polls [2] from our review of R2 and for each determine if
> there is new information that should lead us to re-poll those
> questions. Of course, discussion may lead us to re-polling those if we
> determine that significant new information exists any way.

Poll 1: I have improved the wording and meaning of the byte interface to
allay concerns.

Poll 2: We now match filesystem::path exactly.

Poll 3: No char32_t interfaces, others including a standard library
implementer thought it pointless as there is no code anybody can think
of which consumes a char32_t encoded filesystem path.

> 3) Consider any new polls you might like to take for R3.

My personal biggest concern is with .compare<>(). R3 does not describe
the Unicode implementation corner case choices I eventually chose when
writing the implementation, so I would personally find it valuable if
SG16 reviewed each of those choices I made for wisdom.

Niall