search

cplusplus / papers

ISO/IEC JTC1 SC22 WG21 paper scheduling and management
612 stars 19 forks source link

P2636 References to ranges should always be viewable #1302

Closed wg21bot closed 1 year ago

wg21bot commented 1 year ago

P2636R0 References to ranges should always be viewable (Hannes Hauswedell)

brycelelbach commented 1 year ago

SG9 should see this first.

wg21bot commented 1 year ago

P2636R1 References to ranges should always be viewable (Hannes Hauswedell)

brycelelbach commented 1 year ago

@inbal2l label as Library Evolution when this is ready for us.

brycelelbach commented 1 year ago

Reviewed by SG9 at Kona2022 meeting (Full Minutes).

Polls

POLL: We support applying the resolution for “DE#472: DE-115 26.7.6.1 [range.all.general], 26.4.5 [range.refinements] copyable and movable views” (Apply P2636R1) for C++23.

SF F N A SA
1 0 2 5 2

Attendance: 18 (10 on-site, 8 remotely)

# of Authors: 1

Author’s position: SF

Outcome: Consensus Against

Summary

We had a broad discussion, on which we've seen examples of the issue (unexpected behavior for some of the ranges, which does not qualify as views).

The poll results suggest that SG9 Does not support applying the change proposed. The main concern was that the change conflicted with the fundamental attributes of views.

The issue was forwarded to LEWG.

brycelelbach commented 1 year ago

2022-11-11 08:00 to 09:45 UTC-10 Kona Library Evolution Meeting

P2636R1: References to ranges should always be viewable

2022-11-11 08:00 to 09:45 UTC-10 Kona Library Evolution Minutes

Champion: Hannes Hauswedell (in-person)

Chair: Bryce Adelstein Lelbach & Inbal Levi

Minute Taker: Guy Davidson

Start: 2022-11-11 08:14 UTC-10

Does this paper have:

The wording in the paper is not readable in dark mode.

An alternative solution to this problem would be to have a views::ref or some similar adaptor that wraps move-only views into ref_view and pass everything else through.

POLL: Send P2636R0 (References to ranges should be viewable), the proposed resolution to DE-115, to Library for C++23 classified as B2 - improvement, to be confirmed by a Library Evolution electronic poll.

Strongly Favor Weakly Favor Neutral Weakly Against Strongly Against
1 5 5 8 3

Attendance: 16 (in-person) + 11 (remote)

# of Authors: 1

Author Position: SF

Outcome: No consensus.

POLL: We want a solution to the problem described in DE-115 / P2636R0 (References to ranges should be viewable).

Strongly Favor Weakly Favor Neutral Weakly Against Strongly Against
6 7 6 3 0

Attendance: 16 (in-person) + 11 (remote)

# of Authors: 1

Author Position: SF

Outcome: No consensus.

__POLL: Add an adaptor that wraps move-only views into ref_view and passes everything else through.__

Strongly Favor Weakly Favor Neutral Weakly Against Strongly Against
1 7 4 2 0

Attendance: 16 (in-person) + 11 (remote)

# of Authors: 1

Author Position: SF

Outcome: No consensus.

WA: I would be happy to vote for this if there were other compelling use cases. I didn't find this use case compelling.

End: 09:23

Next Steps

The authors should seek new information or new solutions that will improve consensus. If we don't have any new information by the 2023-02 meeting, we will resolve this NB comment as no consensus for change.

wg21bot commented 1 year ago

P2636R2 References to ranges should always be viewable (Hannes Hauswedell)

h-2 commented 1 year ago

The authors should seek new information or new solutions that will improve consensus. If we don't have any new information by the 2023-02 meeting, we will resolve this NB comment as no consensus for change.

A draft of R2 was discussed by SG9, but did not improve consensus. At this point, I don't think LEWG needs to spend time discussing it again.

brycelelbach commented 1 year ago

This paper is remanded to the custody of the Ranges Study Group. Please either reach consensus on a direction or close this.

inbal2l commented 1 year ago

This was already re-reviewed by SG9 on January 2nd (Minutes) The authors' comment was posted after the second discussion.

In Kona, LEWG has asked that the author will address the comments and update the paper, and then have an additional discussion on SG9, to see if opinions have changed. From reading the room (and emails), the opinions haven’t shifted.

The NB comment (and paper) are passed back to LEWG with the status “no change in SG9's opinion”. Up to LEWG to decide if they want to see it again or close the NB comment as "rejected".

Please also notice the related NB comment' status: https://github.com/cplusplus/nbballot/issues/472

brycelelbach commented 1 year ago

I got this mixed up with another paper. We'll look at it.

brycelelbach commented 1 year ago

Closing with the author's consent, as they no longer wish to pursue this.