Thomas de Zeeuw

> FWIW, a (very basic, FDs/PROC only) implementation of `kqueue` is being worked on and should make its way into Haiku nightly builds before too long, hopefully removing the need...

Thanks @Janrupf for the implementation!

@Noah-Kennedy is this still something you're working on?

Then I'm closing this for now.

Looking at https://cirrus-ci.com/task/5675194417676288?logs=amd64_test#L128 this is still a problem. But since it's just in testing and seems to depend on OS settings (?) given https://github.com/tokio-rs/mio/issues/959#issuecomment-496000849, I'm going to remove this from...

I've been working on io_uring in a different repo: https://github.com/Thomasdezeeuw/a10. Maybe it can become a Mio v2, maybe it should separate as it doesn't support anything other than Linux at...

The original idea of the policy sounded good. But the reality turned out different. The reality is that I maintain most of Mio myself and ping people that are knowledgeable...

> At `polling`, we're trying to do the same thing, but we've run into a problem. If you `register()` a `Source` but then drop it before `deregister()`ing it, a dangling...

@notgull > Unfortunately `kqueue` keeps track of resources through their file descriptor rather than their file description, meaning that the `fd` is morally borrowed by the API for the duration...

> which has the added side effect that we can't deregister it to fix the problem in that case also. > > but overall I don't think the differences between...