char_traits<char8_t> members not all constexpr

[BoPersson]

The paper P1032 "Misc constexpr bits" added constexpr to all char_traits members, at the same time as P0432 "char8_t: A type for UTF-8 characters and strings" added a new specialization for char_traits<char8_t> (without constexpr).

Depending on the order of applying the papers we could get two different results.

The intention of P1032 surely is that even the new members should be constexpr.

[jwakely]

I agree. Arguably, the intention is clear, so it could be considered editorial, but it might need to go through LWG.

[jensmaurer]

@zygoloid, is this editorial?

[tahonermann]

My intent was certainly to keep the char8_t specialization consistent with the others. I agree with fixing this editorially if @zygoloid is comfortable with that.

[zygoloid]

OK, I've reviewed the history here. As noted, P1032R1 and P0482R6 were moved at the same meeting (and P0482R6 was moved first); P1032R1 says to apply a pattern systematically across all char_traits specializations, rather than calling out the specific existing specializations. So I think we merged the applications of P0482R6 and P1032R1 inappropriately; the right way to combine them would have been to make these three functions constexpr for char8_t.

Let's make this change editorially in the name of fixing the misapplication of prior motions, but send a notification to LWG to inform them that we're doing so.

[Dani-Hub]

I noticed the problem described in this issue today and was considering to open an LWG issue for this. Can I relax now and assume that an editorial change will be applied instead?

[jensmaurer]

LWG chair was informed:

Subject: [isocpp-edit] [char.traits.specializations.char8.t] Make all members constexpr
Date: Mon, 27 May 2019 09:07:07 +0200
From: Jens Maurer Jens.Maurer@gmx.net
Reply-To: IS and TS editorial discussions edit@lists.isocpp.org
To: Marshall Clow mclow.lists@gmail.com
CC: IS and TS editorial discussions edit@lists.isocpp.org

Dear LWG chair:

This is to inform you that an editorial change has been applied
that you might consider a non-trivial merge of two papers
approved at the 2018-11 meeting:

#2833

Thank you.
Jens (co-editor)