Slides for P3793R0 — Better shifting

Document number:: P3898R0
Date:: 2025-11-02
Audience:: SG6
Project:: ISO/IEC 14882 Programming Languages — C++, ISO/IEC JTC1/SC22/WG21
Reply-to:: Jan Schultke <janschultke@gmail.com>
Source:: github.com/Eisenwave/cpp-proposals/blob/master/src/better-shifting-slides.cow

This document has custom controls:

→, ↓ : go to the next slide
←, ↑ : go to previous slide

Better shifting
P3793R0

Brian Bi, Jan Schultke | Slides for P3793R0 — Better shifting | Kona 2025 | Slide 1

Introduction

bit-shifting (<< and >>) is not ideal:
- hard-to-remember precedence rules
- undefined behavior on "overlong" shifts
- other languages, hardware use modular shift amount (C#, Java, JS)
"fixing" bit-shifts in core language difficult
- at best, unspecified result + erroneous behavior
proposed: "mathematically correct" std::shl and std::shr

Brian Bi, Jan Schultke | Slides for P3793R0 — Better shifting | Kona 2025 | Slide 2

Motivation

"Mathematically correct" functions eliminate special cases and UB pitfalls:

Before	After
uint32_t make_mask(int n_bits) { if (n_bits >= 32) return uint32_t(-1); return (uint32_t(1) << n_bits) - 1; }	uint32_t make_mask(int n_bits) { return shl(uint32_t(1), n_bits) - 1; }
uint32_t bitset = /* ... */; bool bitset_contains(size_t i) { if (i >= 32) return false; return (bitset >> i) & 1; }	uint32_t bitset = /* ... */; bool bitset_contains(size_t i) { return shr(bitset, i) & 1; }

Brian Bi, Jan Schultke | Slides for P3793R0 — Better shifting | Kona 2025 | Slide 3

Design choices

std::shl or std::shr N-bit shift acts like N single-bit shifts
no wrapping behavior because it's not useful
no separate logical/arithmetic right-shift (change signedness to choose)
negative shifts have erroneous behavior with impl-defined result
- no added runtime cost on release builds, no UB 🎉
function names inspired by std::rotl and std::rotr
shift amount passed as int (like std::rotl and std::rotr)
SIMD support added, similar to std::simd::rotl

Brian Bi, Jan Schultke | Slides for P3793R0 — Better shifting | Kona 2025 | Slide 4

Possible implementation

template

signed-or-unsigned

shl

(

int

)

noexcept

{

return

unsigned

(

)

width-v

(

)

unsigned

(

)

;

}

template

signed-or-unsigned

shr

(

int

)

noexcept

{

return

unsigned

(

)

width-v

(

)

unsigned

(

)

;

}

Brian Bi, Jan Schultke | Slides for P3793R0 — Better shifting | Kona 2025 | Slide 5

References

[P3724R1] Jan Schultke. Integer division 2025-09-29 https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2025/p3724r1.html

[GitHub] Jan Schultke. integer-division GitHub repository https://github.com/Eisenwave/integer-division

[StackOverflowCeil] Fast ceiling of an integer division in C / C++ https://stackoverflow.com/q/2745074/5740428

Slides for P3793R0 — Better shifting

Better shifting P3793R0

Introduction

Motivation

Design choices

Possible implementation

References

Better shifting
P3793R0