P3371R1 Fix C++26 by making the rank-1, rank-2, rank-k, and rank-2k updates consistent with the BLAS (Mark Hoemmen, Ilya Burylov)

P3371R2 Fix C++26 by making the rank-1, rank-2, rank-k, and rank-2k updates consistent with the BLAS (Mark Hoemmen, Ilya Burylov)

P3371R3 Fix C++26 by making the rank-1, rank-2, rank-k, and rank-2k updates consistent with the BLAS (Mark Hoemmen, Ilya Burylov)

2025-03-18 Library Evolution Telecon

P3371R3: Fix C++26 by making the rank-1, rank-2, rank-k, and rank-2k updates consistent with the BLAS

2025-03-18 Library Evolution Hagenberg Minutes

Champion: Mark Hoemmen / Ilya Burylov
Chair: Inbal Levi
Minute Taker: Bronek Kozicki

Summary

POLL: We are unhappy with the behaviour currently specified in the paper “Use only the real part of scaling factor alpha for Hermitian matrix rank-1 and rank-k updates”.

Attendance: 17
Author's Position: SA
Outcome: Consensus against

POLL: Reverse the change for “Remove rank-{1,k} overloads without a scaling factor alpha (keep overloads with alpha)”
(voting in favor would mean keeping the overloads)

Attendance: 18
Author's Position: A
Outcome: No consensus for a change

IL: We have some comments on the poll (in respect to moving away from status quo), so just to make sure, let’s take the poll again in the opposite direction (approving modifications proposed in the paper).

POLL: Apply the change as proposed in the paper “Remove rank-{1,k} overloads without a scaling factor alpha”.

Attendance: 18
Author's Position: SF, F
Outcome: Consensus in favor

A: I think the overload might be useful.

POLL: Forward “P3371R3: Fix C++26 by making the rank-1, rank-2, rank-k, and rank-2k updates consistent with the BLAS” to LWG for C++29 with a recommendation to make this a DR for 26.

Attendance: 17
Author's Position: 2xSF
Outcome: Strong consensus in favor

POLL: Forward “P3371R3: Fix C++26 by making the rank-1, rank-2, rank-k, and rank-2k updates consistent with the BLAS” to LWG with a recommendation to apply for C++26 (if possible).

Attendance: 17
Author's Position: 2xSF
Outcome: Strong consensus in favor

Next Steps

Paper was forwarded for C++29 with a recommendation for making this DR for 26. As this is a fix, we also took a second (non-binding) poll that recommends accepting this for C++26 (if LWG’s queue enables).

P3371R4 Fix C++26 by making the rank-1, rank-2, rank-k, and rank-2k updates consistent with the BLAS (Mark Hoemmen, Ilya Burylov)