Announcing Rust 1.29 | Hacker News Reader

Announcing Rust 1.29

blog.rust-lang.org

108 pointssteveklabnik10 comments

    cargo doc --document-private-items

This is going to be excellent. I often find in Rust, more than other languages, that the documentation really helps fill a mental gap around the types that you expect a type to implement, vs. what it actually implements. The compiler always catches this, but the docs are also a great way of discovering this.

It's going to be a huge help when working on internal interfaces in a library. cargo doc is such an excellent tool, and while the UI took a minute to learn when I first used it, it's really an amazingly helpful tool, and feels more integrated than similar tools in other languages.

A note to any Rust maintainers reading:

There is something wrong with the line spacing. It's so narrow that it's hiding the underscore in one of the code examples:

https://screenshots.firefox.com/PYoTEXQwowBJoG30/blog.rust-l...

Firefox 62 on Linux (Ubuntu 16.04)

Thank you! I recently merged a PR to change some CSS; I'll tweak and/or revert it. It looks okay on Firefox on Windows...

https://github.com/rust-lang/blog.rust-lang.org/issues/269

Glad to hear that 1.29 now brings clippy to stable Rust. (Before, clippy required nightly Rust.)

What's the status for futures/async-await? Or, asking more generally: Where can I go to keep myself up-to-date with this topic?

Status is, there's an initial, incomplete implementation in nightly. It's not going to land in stable until early next year.

https://github.com/rust-lang/rust/issues/50547 is the tracking issue.

A question about Cargo Fix. Is that only on warnings, or can/does it currently fix errors it knows the problem with as well?

I'm not 100% sure, but right now, the focus has mostly been on the use-case of upgrading to Rust 2018, which is warnings-based. I'd imagine it would eventually support both, but am not 100% sure about where exactly it is today.

Errors are a bit more challenging, because fixing an error may introduce more, new errors, making it harder to tell if applying the fix is correct or not.

That's part of why I was wondering. Looking forward to playing with it (plus having Clippy on stable since that's all I use)

I asked the dev on Discord and he said:

> I think in theory we could fix errors too, because they have the same diagnostics format. Don't think we do that right now though, and we are probably not prepared to handle the compiler exiting with a non-zero code

So yeah, seems like that's the issue.

Reply on news.ycombinator.com