Nemo157

> Please also note that selecting on `stream.next()` is a very problematic thing to do, as it makes a fresh new `.next()` call on each iteration and then drops (cancels)...

There is explicit documentation in `std` now that returning a self-constructed `std::fmt::Error` is an "incorrect implementation": https://doc.rust-lang.org/stable/std/macro.format.html#panics

> You can write a standalone linker script Yep, that's my current plan. Unfortunately you cannot have multiple `SECTIONS` defined in different linker scripts, so that requires copying the full...

Hmm, you may be right. I'm sure I tested this earlier and had it ignore the sections specified in `link.x` when I added some extras, but it looks like it's...

So, to clarify: the behaviour after this PR is that commas at the end of multi-item metalists will not be added/removed at all. If the attribute is converted between multi-line/single-line...

What is necessary to perform the backport here? If it's just cherry-picking the commits onto another branch I'm willing to do that if you let me know which branch.

> (You run a resolver locally?) `systemd-resolved` is relatively common and runs a stub resolver on 127.0.0.53:53 (though I see no way to make it listen on IPv6).

Also, very useful in cases where you query multiple nameservers, per-line so you can see which nameserver returned each result: ```console > dog @1.1.1.1 @134.195.4.2 A google.com A google.com. 3m41s...

> although we'd have to pre-build all the docs for the provided crates like we pre-compile them. My expectation would be that it would be like `cargo docs --no-deps`, linking...

> is there a way to get it to point to docs.rs? There's the `--extern-html-root-url NAME=URL` flag for `rustdoc` which I believe can be used to override the `html_root_url` set...