Louis Wasserman

We've been thinking about this some at Google, especially based on Guava's experience with RateLimiters, which has gotten some negative feedback for its inability to burst or to manage bursts....

Continuing conversation from https://github.com/Kotlin/kotlinx.coroutines/pull/3278#issuecomment-1127402291. ## We don't mind if it's slow I've mentioned this in other contexts, but perhaps not when you were present, so let me say this as...

(To clarify, in the "parsing streams of bytes" use case, we are returning a `Flow` of the parsed messages.)

I don't have specific issues I can refer to about Unconfined or UNDISPATCHED, though both make us extremely nervous. Certainly in Java land we've had lots of issues caused by...

Does that approach address the race issue discussed in the PR [here](https://github.com/Kotlin/kotlinx.coroutines/pull/3278#issuecomment-1125513103)?

There are plenty of races that can happen as a result of coroutines and not threading! https://github.com/Kotlin/kotlinx.coroutines/pull/3278#issuecomment-1124141560 is an example of code that nondeterministically throws or succeeds, even run on...

I agree that these are two different sorts of problem, but I'm not actually convinced of your relative ordering of them. Here are some counterpoints I'd like to make: *...

I agree that there's no way to make _any_ coroutines code "conceptually deterministic," but I certainly claim that you can have deterministic tests for many coroutine tests, especially at the...

(Not ready for submission; needs tests.)

We want something linear: something that doesn't buffer ahead, something that doesn't collect the flow concurrently with the handling of its elements. This is especially bad in the case of...