[2026-06 LWG Motion 11] P3793R2 Better shifting

[eisenwave]

Fixes #9098
Also fixes cplusplus/papers#2399

Note

The rename of simd-vec-type to simd-integral was deliberately omitted for now so that we can have a green CI build. It's trivial for Thomas to apply that or for a fixup commit to add that in.

[eisenwave]

@jwakely @brevzin last chance to deal with
#9109 (comment)

Both these papers bump __cpp_lib_bitops but LWG wanted there to be distinct bumps, so does that mean 202606L/202607L, or does it mean 202605L/202606L?

We also have the same situation for __cpp_lib_simd_bitops, which is introduced in this paper, but then immediately bumped by motion 15.

[jwakely]

@jwakely @brevzin last chance to deal with #9109 (comment)

Both these papers bump __cpp_lib_bitops but LWG wanted there to be distinct bumps, so does that mean 202606L/202607L, or does it mean 202605L/202606L?

It doesn't matter.

I thought we did it as increasing from the date of the meeting but __cpp_lib_format does it the other way. I don't care, it doesn't matter.

[eisenwave]

@tkoeppe still gotta replace simd-vec-type with simd-integral.

[tkoeppe]

Thanks. Does the paper mis-cite the signatures of rotl and rotr`?

[eisenwave]

Thanks. Does the paper mis-cite the signatures of rotl and rotr`?

It might be citing the already applied motion from another paper that replaced simd-vec-type with simd-integral for those other functions.