Compat blog post take 2

[Nemo157]

Replacement for #1226 (@MajorBreakfast I hope you're ok with me taking this over).

I still want to add some info on the spawn and IO compat that's been added since this was originally written, and I need to actually test the snippets against the new API. Should get to that over the weekend at some point.

[cramertj]

Aside from the one nit and updating to add IO/spawn info, this LGTM so far!

[LucioFranco]

One, thing that might be useful to add is something about sink support with tokio. As I used this post to convert my project to async/await. I was a bit confused on how to approach sinks. I ended up with using some of the tokio compat features. So maybe a mention about it would be helpful imo.

[Nemo157]

@LucioFranco yeah, probably worth adding a section about Sink. Currently it's possible to wrap a 0.3 Sink into a 0.1 Sink, but not the other direction. I will also open an issue about whether it's possible to add that additional conversion, I couldn't think of an easy way when implementing the current Sink compat stuff but maybe it's possible with a little bit of extra runtime cost.

[Nemo157]

Added info on Spawn, IO and Sink, and updated the final example to use the new IO compat layer.

[LucioFranco]

@Nemo157 Great! I just took a look at it again, it def clears some things up. I am still slightly struggling to understand where the TokioDefaultSpawner fits into the whole picture. Also, I think the fact that there is no 0.1 -> 0.3 is a bigger issue for people using tokio. So maybe might be worthwhile to be more upfront about that?

This btw is my hacky solution to using std await and UdpFramed from tokio. I would be interested to help build out some of the sink compat stuff.

while let Some(msg) = await!(rx.next()) {
    println!("Sending: {:?}", msg);

    {
        use tokio::prelude::SinkExt;
        if let Err(e) = await!(sink.send_async((msg, to))) {
            println!("Sending message: {}", e);
        }
    }
}

I would also maybe add a cargo.toml snippet for the example so that its easier for people to try it out.

[Nemo157]

@LucioFranco thanks, those are some good points. I think an example of using TokioDefaultSpawner is probably the easiest way to show what it’s useful for. I can expand a bit more on the Sink situation as well. And having the Cargo.toml for the example would definitely be worth it.

[Nemo157]

Sorry about the big delay on this again, I'll try and finish this off before 2019!

[Nemo157]

Updated now that Sink is supported and TokioDefaultSpawner was removed, also uses the new methods from #1469 for consistency so is blocked on that being merged and a new release.

[LucioFranco]

Read through it again looks good!

[corbinu]

@Nemo157 Read through it as somebody outside of the working team and just excited about async/await. It looks great and is easy to understand. Super excited!

[archseer]

How would one go about using select! with tokio compatibility? I ran into some problems because select-macro seems to call out to the stdlib/futures implementations so it doesn't hit the tokio await wrapper.

[Nemo157]

@archseer I have no real experience using Tokio's await! macro, but you can just use the futures 0.1 -> 0.3 compat methods to convert Tokio's futures into std futures so that they work with std::await!, something like

use futures::future::FutureExt;
use futures::compat::Future01CompatExt;

select! {
        foo = some_tokio_future().compat().fuse() => { ... },
        bar = some_other_tokio_future().compat().fuse() => { ... },
};

[Nemo157]

@cramertj I've checked the example code works with the new release and all the links are correct, I think this is good to publish finally 😁

[cramertj]

Nice!