Add LocalWaker and ContextBuilder types to core, and LocalWake trait to alloc.

[tvallotton]

Implementation for #118959.

[rustbot]

Thanks for the pull request, and welcome! The Rust team is excited to review your changes, and you should hear from @Mark-Simulacrum (or someone else) soon.

Please see the contribution instructions for more information. Namely, in order to ensure the minimum review times lag, PR authors and assigned reviewers should ensure that the review label (S-waiting-on-review and S-waiting-on-author) stays updated, invoking these commands when appropriate:

• @rustbot author: the review is finished, PR author should check the comments and take action accordingly
• @rustbot review: the author is ready for a review, this PR will be queued again in the reviewer's queue

[tvallotton]

This CI failure is strange:

Error: tidy error: /checkout/library/core/src/task/wake.rs:713: issue "87021" mismatches the previous issue of "96992"

It seems the error message is correct about the fact that we shouldn't be linking to #87021 and we should link instead to #96992. But it makes me question why does it only fail on my branch, I didn't change this issue number after all.

[rustbot]

rust-analyzer is developed in its own repository. If possible, consider making this change to rust-lang/rust-analyzer instead.

cc @rust-lang/rust-analyzer

Just for the record, it might be wise to check if ContextBuilder<'a> should be future proofed to be invariant, since I noticed that the with_waker example seems to break when changing the variance. Naturally, that makes the implementation of From<&mut Context> practically useless.

EDIT: This implementation was later removed

Do we have signoff from T-libs-api on the new unstable surface area? The ACP issue seems to have had a good bit of discussion but I didn't see that signoff made explicit. If you can drive that towards a comment here or there and/or link something, that would be good.

I'm particularly concerned about the surface area confusing newer users -- especially given the frequent concerns over async complexity -- I wonder if we should add a sentence/paragraph to the top-level docs in https://doc.rust-lang.org/nightly/std/task/index.html to clarify the current state (i.e., stable surface area should be sufficient for most uses or so).

I wonder if we should add a sentence/paragraph to the top-level docs in https://doc.rust-lang.org/nightly/std/task/index.html to clarify the current state (i.e., stable surface area should be sufficient for most uses or so).

I think it may be better to put it on the docs for LocalWaker. What do you think?

I like to see a LocalWaker type, but this not the way to do it in my opinion. That's coming from someone who has written a lot of Futures, async function and a runtime specialised in thread-locality (i.e. I very much like to see this change).

I don't think making the Waker optional in Context is wise considering all existing Future related code depends on the Waker type always being part of the Context, it was never not there, a promise we're walking back with this pr.

I made the waker type mandatory on the last commit. Several people have expressed their concern with this aspect of the API. And while I do think that it is better for them to be optional than not, that is not something I consider terribly important.

☔ The latest upstream changes (presumably #120136) made this pull request unmergeable. Please resolve the merge conflicts.

@Mark-Simulacrum Do we really need an explicit signoff here? Both teams have discussed it and approved nightly experimentation (libs-team, wg-async). I feel that with no significant changes on the proposal, it is going to be hard to get the teams to meet to discuss the same topic again.

Those comments seem sufficient to me (I hadn't seen them in my quick skim) for nightly experimentation.

@rustbot ready

@bors r+

📌 Commit 180c68b has been approved by Mark-Simulacrum

It is now in the queue for this repository.