Shared stream

[tmccombs]

This is similar to #1217 but uses an underlying boxed slice to buffer items, and as a side-effect has a finite buffer size.

This is a work in progress. I'd still like to add some more rigorous unit tests, but I'd like feedback on my general approach, and have other eyes check my logic.

[ebkalderon]

Has there been any additional progress on this recently? I happen to be looking for this exact feature.

[tmccombs]

I haven't done anything more with this. I would like some feedback on my approach before I put more effort into polishing it.

[tmccombs]

@cramertj, @MajorBreakfast any thoughts on this?

[ebkalderon]

Is this PR still being worked on? I ran into several more cases over the past two years where this API could be useful. I'm willing to take over this PR and help drive it to completion if the original author is away. @cramertj, any thoughts on this?

[tmccombs]

I'm still around. It's been a while since I looked at this, but I was waiting on feedback from the maintainers.

[tmccombs]

Looks like the shared future::Shared was refactored a little bit. So I rebased my branch on master. One piece that is still missing from this is some unit tests for stream::Shared. If you wanted to help write some @ebkalderon that might be helpful.

[ebkalderon]

Sure thing, @tmccombs! I'll write a few, perhaps based on the tests for future::Shared, and I'll post them here.

[ebkalderon]

@tmccombs I didn't get around to writing tests for this, unfortunately, but it seems that the current commit is failing in CI due to a clippy lint for unnested-or-patterns (source). I can write some tests once this is fixed and open a PR against your branch, perhaps, so you can merge them in and have them ready for review here.

[tmccombs]

@ebkalderon those clippy lints are for code I didn't touch, and making the suggested change results in an "or-patterns syntax is experimental" error on stable rust.

[ebkalderon]

Got it! Thanks for clarifying and updating the branch. I'll get started writing some tests and open a PR against your fork.

[tmccombs]

Conflicts have been resolved and this is up-to-date with current master.

[tmccombs]

The failed miri test looks like it is probably unrelated, but I'm not completely sure.

[XLordalX]

Any updates?

[tmccombs]

@XLordalX your comment got me to revisit this, and I wrote some tests, which led to finding some bugs which I have now fixed, and reworked it a fair amount.

It is still failing CI. Something about a race condition found with thread sanitization. I don't really understand it yet.

I would kind of like some feedback from the futures maintainers to know if this a direction worth pursuing. And, to be honest, I kind of hope someone comes along and says there is a simpler way to do it.

[XLordalX]

@taiki-e You able to look at this?

[tmccombs]

hmm, looks like builds are currently broken on master.

[rustbot]

This PR was rebased onto a different master commit. Here's a range-diff highlighting what actually changed.

Rebasing is a normal part of keeping PRs up to date, so no action is needed—this note is just to help reviewers.