P3306 R0 Atomic Read-Modify-Write Improvements

[wg21bot]

P3306R0 Atomic Read-Modify-Write Improvements (Gonzalo Brito Gadeschi, Damien Lebrun-Grandie)

[mattkretz]

P3306R0: Atomic Read-Modify-Write Improvements

2024-06-24 SG6 Minutes

Chair: Matthias Kretz

Champion: Damien Lebrun-Grandie

Minute Taker: Frank Birbacher

Summary

There was a concern that the name "nand" is confusing; and that the function is
unnecessary if it implements an "and-not" - if anything, it needs to be a "not-and". The
paper should add motivation for why a "not-and" is useful.

Also the paper should spell out how operations on floating-point resulting in
NaN work. The expectation in SG6 was bit-wise comparison, rather than
floating-point comparison (in a CAS loop).

SG6 sees no need to see the next revision of this paper. There is no real
numerics concern.

[ogiroux]

SG1 in St Louis

No objection to more work on D3306R1 based on the minuted feedback provided