Filip Gospodinov

The tests fail for me also without this commit :man_shrugging:

> The reason we use `___foo` instead of `foo` or `r#foo` is because of the lack of hygiene for proc-macros. Thanks for providing some context here. Given that those variables...

I can recommend taking a look at [axum](https://docs.rs/axum/0.2.5/axum/index.html) to those that require built-in websocket support (needs to be enabled via a [feature flag](https://docs.rs/axum/0.2.5/axum/index.html#feature-flags)). It has similar usage ergonomics like Rocket...

@OvermindDL1 I don't think there's anything inappropriate about my post. I'm a rocket user myself but I cannot use it in services where I need built-in websocket support. Axum is...

@OvermindDL1 interesting! I've heard about axum just recently. That's why I felt it's worth sharing about it after I experienced how well designed it is. > What would be nice...

@OvermindDL1 that is exactly my use-case and I had also to reach for axum to make it work :) With that said, I feel it's slightly off-topic to discuss ideas...

> We don't consider these bugs Agreed, it's technically not a bug. But it's terrible macro hygiene. Your generated code shouldn't trigger lints in my crates.

My apologies if this came across condescending. That's not the intention. I don't agree with your statement but I also think we both have made our points already.

This crate could easily prevent such breaking changes by re-exporting crates that are used by its public API such as `rand`. I made a PR to fix this https://github.com/dalek-cryptography/ed25519-dalek/pull/159. Well-maintained...

Agreed, the user can always shoot herself in the foot. There's no silver bullet but it would be at least an improvement.