Duarte Nunes

Thanks for the patch! However, seastar doesn't accept pull requests; we use a mailing list instead. See https://github.com/scylladb/seastar/blob/master/CONTRIBUTING.md for instructions.

I have the same issue: it works fine when using otel's stdout exporter, but the root span is missing when using an `opentelemetry_otlp` exporter (which suggests an issue on the...

This makes sense to me too. Since errors are already injected in the response, an empty payload object can be a simple indicator of success. It's less confusing and wasteful...

I like the idea! Some small comments.

Currently working on this. Going to initially focus on successful compilation, then execution without the ability to do any I/O, and eventually fallback to native, synchronous I/O.

My assumption was that calling synchronous syscalls was significantly less effort than using `kqueue` or something, and that would at least unblock developing on the the mac. I'll give it...

I'm a bit confused because the PR just seems to manually writes out the future that `async` would generate. You'll also need to wrap the resulting future in a `Box`...

> **Except that it's an async function**: I can't use it directly from `poll_next` (a poll function is obviously not `async`). > Why can't you call an `async` function from...

> Greeting @duarten! Howdy! :) > > > Why can't you call an async function from poll_next? > > You can, but you can't `.await` it and you need to...

> @duarten your example works because you use the stack for storage. Try to make this work with a stream, and you'll realize you can't always do that. Hum.. not...