Add async example to bindgen_examples

[ifsheldon]

Closes #9776

I reused most of example 4 to avoid extra complication. Also adds a reference to code example wasi_async, so it'd better if this gets merged after #9788.

cc @alexcrichton

[alexcrichton]

Thanks, and looks good to me! I think there's some errors on CI though?

[ifsheldon]

Can you take a look into the clippy error please? The error seems to originate from bindgen!. I suspect that some code in bindgen! forgets to import Box when a WIT spec is inlined. In my own example that imports from a WIT file, there's no problem. In the example 4 where async is not used, it's also fine. I have no clue why that happened. The difference between _7_async.rs and _4_imported_resources.rs is just the line async: true,

[alexcrichton]

Ah I believe this is exposing a preexisting bug in the bindgen! macro, notably the async output is not compatible with #![no_std] crates. The wasmtime crate itself is such a crate which is where this is coming from.

I'm not sure of a way to work around this other than fixing the issue at hand (tracking down the Box and importing it from wasmtime::component::__internal::Box)

[ifsheldon]

I think for the purpose of documentation, it's okay to let clippy skip this file.

this is exposing a preexisting bug in the bindgen! macro, notably the async output is not compatible with #![no_std] crates. The wasmtime crate itself is such a crate which is where this is coming from.

IIUC, #9823 maybe unblock this?

I'm not sure of a way to work around this other than fixing the issue at hand

Can you try to fix this please? I'm not very familiar proc_macros or bindgen internals.

[alexcrichton]

Yes #9823 would probably help. Unfortunately I don't have time right now to fix this myself, and this can't land unless it's passing CI.

[ifsheldon]

OK...... Can you find the issue tracking the "preexisting bug" you mentioned? I can't find one to link here. Or, I can file a new one.

[alexcrichton]

I don't believe there's a preexisting issue since this is I think the first time it's come up, but feel free to file an issue of course

[ifsheldon]

OK, let's take a step back. Why need to build bindgen_examples inside wasmtime anyway? I think we can just move all the bindgen examples out of wasmtime crate.

[alexcrichton]

It's not necessarily strictly required, but it's quite useful to be distributed as part of docs.rs integration. That gives nice API docs all in one place. Personally I think it'd probably be less effort to fix the code generation than it would be to move all the documentation somewhere else.

[ifsheldon]

OK, fine. I made a new commit which fixes the bug of bindgen. I know I should have open another PR/issue but it seems I need to provide an example to reproduce the bug, which is this very example, so I think it's okay to just fix it in this PR.

[alexcrichton]

Thanks! I think some test expectations may need an update, but otherwise looks good 👍