[IN PROGRESS] Update to mio 0.7 and tokio 0.3

[witchof0x20]

This is my attempt at updating to mio 0.7, as referenced in #6

This is likely not entirely correct, and I plan to go back over it some time this week. Putting it up as a PR to make it easier to see changes.

[witchof0x20]

Tested this using
main.rs

use futures::executor;
use futures::stream::StreamExt;
use std::thread;
use tokio_udev::{EventType, MonitorBuilder};

#[tokio::main]
async fn main() {
    let mut monitor = MonitorBuilder::new()
        .unwrap()
        .match_subsystem("usb")
        .unwrap()
        .listen()
        .unwrap();

    loop {
        if let Some(event) = monitor.next().await {
            println!("{}", event.event_type());
            for property in event.device().properties() {
                println!("{:?} = {:?}", property.name(), property.value());
            }
        }
    }
}

Cargo.toml:

futures = "0.3"
tokio-udev = {path = "../tokio-udev/tokio-udev"}
tokio = { version = "0.3", features = ["full"] }

and it works great

With tokio-udev from cargo, I actually get
thread 'main' panicked at 'there is no reactor running, must be called from the context of Tokio runtime'
with this code. I haven't dug into making my example code work with instantiating the reactor yet.

I did some additional testing, and it seems this PR would also allow for bumping this project's tokio dependency to 0.3. There are no compilation errors, and the example code works.

[witchof0x20]

https://github.com/jeandudey/tokio-udev/blob/master/tokio-udev/examples/usb_hotplug.rs
works with this PR
It doesn't work with tokio 0.3 (not in this PR), but that's for a separate PR I think

[jeandudey]

@witchof0x20 did it not work with udev-rs@0.5? To merge #10 first and then merge the tokio update code (when done).

[witchof0x20]

@witchof0x20 did it not work with udev-rs@0.5? To merge #10 first and then merge the tokio update code (when done).

It works with udev-rs@0.5, I just split it up because that one is an easy merge, while this is a bit more complicated.

[witchof0x20]

Getting back to this, as I'm more free now. As I understand it, I would modify tokio_udev::Inner such that it wraps AsyncFd<MonitorSocket> (with an AsRawFd trait implementation that's just a wrapper for MonitorSocket::fd).

Then, AsyncFd::poll_read_ready can be used to poll the MonitorSocket for availability, pulling events out of it if possible. But that requires a &mut core::task::Context, and I'm not sure where to grab that from, as I'm not super familiar with how modern tokio works anymore.

[witchof0x20]

I think I got this working with AsyncFd. The example compiles and runs, with some changes to be compatible with updated tokio.

Requesting review from @jeandudey

[jeandudey]

Thanks for your contribution! Will add the changes to the CHANGELOG.md and release a new version as soon as possible.

[jeandudey]

Released 0.4.0 of both crates with recent changes and 0.5.0 with udev 0.5