[Proposal]: Unsigned right shift operator

[AlekseyTs]

Unsigned right shift operator

• Proposed
• Prototype: Not Started
• Implementation: Not Started
• Specification: Not Started

Speclet: https://github.com/dotnet/csharplang/blob/main/proposals/csharp-11.0/unsigned-right-shift-operator.md

Summary

An unsigned right shift operator will be supported by C# as a built-in operator (for primitive integral types) and as a user-defined operator.

Motivation

When working with signed integral value, it is not uncommon that you need to shift bits right without replicating the high order bit on each shift. While this can be achieved for primitive integral types with a regular shift operator, a cast to an unsigned type before the shift operation and a cast back after it is required. Within the context of the generic math interfaces the libraries are planning to expose, this is potentially more problematic as the type might not necessary have an unsigned counterpart defined or known upfront by the generic math code, yet an algorithm might rely on ability to perform an unsigned right shift operation.

Detailed design

There are some considerations around what the operator could be and the initial suggestion is >>>, to match other languages. It is suggested that >>> be explicitly an unsigned right shift and >> would continue being just “right shift” with no strict definition. For primitive unsigned integral types >>> will be allowed, but will be equivalent to >>.

Drawbacks

Alternatives

Unresolved questions

Design meetings

[YairHalberstadt]

Is it common enough to need an operator? Couldn't it just be an ordinary static method?

[KalleOlaviNiemitalo]

Roslyn recognizes >>>>>>> as the end of a merge conflict but >>> is shorter and so would not conflict with that.

[333fred]

Roslyn recognizes >>>>>>> as the end of a merge conflict but >>> is shorter and so would not conflict with that.

Also, if we really wanted to use that syntax as an operator, we'd use it. That's just roslyn trying to be helpful, nothing in the C# spec mandates that the compiler produce a good error when encountering merge conflict markers :).

[whoisj]

Also nothing mandates that conflict markers be composed only of > characters. 😉

[BillWagner]

Note: Spec is here