Thomas de Zeeuw
Thomas de Zeeuw
Any thing I can do to help this move forward? Mio v0.6 is no longer supported (and hasn't been for a little while).
@Darksonn my first instinct would be that Tokio isn't checking [`Event::is_error`](https://docs.rs/mio/latest/mio/event/struct.Event.html#method.is_error) when handling read events, but I don't really know where the related code is so it's hard for me...
Looking at [`Ready::from_mio`](https://github.com/tokio-rs/tokio/blob/78e0f0b42a4f7a50f3986f576703e5a3cb473b79/tokio/src/io/driver/ready.rs#L38-L69) it doesn't seem to consider `mio::Error::is_error`. So it already goes wrong in the call here: https://github.com/tokio-rs/tokio/blob/78e0f0b42a4f7a50f3986f576703e5a3cb473b79/tokio/src/io/driver/mod.rs#L174
> What should the fix do? Should `async_fd.readable().await` return an error whenever the `EPOLLERR` happens? Or should it return success (and the user will get the error when attempting a...
> Could we not return an error to the user that basically just says "EPOLLERR happened", and let them retrieve the actual underlying error in the way the find appropriate?...
> This can be traced down to https://github.com/tokio-rs/mio/blob/dca2134ef355b3c0d00e8e338e44e7d9ed63edac/src/sys/unix/selector/epoll.rs#L180-L187 https://github.com/tokio-rs/mio/blob/05009e4f60335fa00e9ea6a118595548afee0607/src/event/event.rs#L116-L123 > > An event with `EPOLLERR` is considered `is_write_closed`, which wakes up awaiters with a write interest. However, `EPOLLERR` does not...
For the named arguments maybe we can use something like [`format`](https://doc.rust-lang.org/std/macro.format.html) does: `format!("x = {}, y = {y}", 10, y = 30);`. As for the language switching my first though...
This https://github.com/microsoft/windows-rs/pull/1636 might be a bit of problem, as Mio supports older rustc versions. Does windows-sys give any MSRV guarantees?
This issue is now blocked on https://github.com/tokio-rs/mio/issues/1562.
Updating to 0.36 in https://github.com/tokio-rs/mio/pull/1569. I couldn't find the `SIO_BSP_*` constants although the linked issue is closed, have those been added @kennykerr?