[Proposal] Remove Sendable conformances from unsafe pointer types

Evolution Pitches
1

I'm reposting a separate mini-proposal: Remove Sendable conformance from unsafe pointer types (proposal PR).

I extracted this proposal from the larger proposal: Improving Sendable Inference.

I didn't see any concern about changing unsafe pointer types in the pitch thread for the larger proposal. And I don't think there is any dependency or close relationship between the other proposed changes:

  • Sendable KeyPath literals
  • Non-Sendable keyword
  • Generic Sendable inference
  • Compile time control over SEndable enforcement

An early discussion of the unsafe pointer change was positive.

17 Likes
2

I'm very much +1 on this narrow change.

Doug

2 Likes
3

Sounds good to me, this only really changes the default automatic behavior to be more safe. And if we need to we can always do unchecked wrappers while actively thinking about it. +1, no concerns really.

4

+1 This seems like a no brainer to me.

5

+1

6

+1 I was astonished to learn that unsafe pointer types were Sendable in the first place.

2 Likes