Sergey Bugaev

> qemu user emulation could work for sure It won't, because we link to native libraries; unless you're suggesting users should install PPC versions of their whole /lib. > If...

"effectively no AppKit support" might just be too strong of an expression — we have tons of AppKit code, and some apps working :slightly_smiling_face: That doesn't mean we have all...

That would not work at all. AppKit has backend(s), and Apple's one is very different from ours. There [has been some work](https://github.com/darlinghq/darling/issues/937) to make our backends at least _look_ more...

> mime type as `application/x-shellscript`, which is not a text mime type, and therefore you cannot paste that script in to text boxes in say, Firefox, for example. For users...

> There should still be a blocking sync API though for use-cases which are simple enough to not need the full async stack (like wl-copy, etc.). That's what `thread::block_on()` is...

`thread::block_on()` itself [should be](https://boats.gitlab.io/blog/post/global-executors/) in std; but you indeed need a reactor.

> Watching the socket requires a reactor; there should be implementations for Tokio and async-std, togglable with a cfg flag. Actually if we already "know" the user is using tokio...

> Yeah, it should be `AsyncRead` (or `Data`) too. Hmm, maybe like this then: ```rust trait ClipboardContents { fn types(&self) -> impl Iterator; fn data(&self, type: MimeType) -> impl AsyncRead;...

My bad, it's [this method](https://docs.rs/tokio/0.2.1/tokio/runtime/struct.Builder.html#method.basic_scheduler) that tells tokio to only use the current thread. And > If the `rt-core` feature is enabled and `rt-threaded` is not, `Runtime::new` will return a...

😀