4 ms·
Rust 2020: A Better Ecosystem
- jedisct1 7y agoWhat I would love is RLS (or rust-analyzer, whatever) that works. Ie. that can properly infer all the types, do refactoring, is fast and stable. When it comes to tooling, moving from Go to Rust is a painful experience. Lifetime visualization is indeed a difficult problem. That blog post brings interesting ideas on this: https://blog.adamant-lang.org/2019/rust-lifetime-visualization-ideas/ https://blog.adamant-lang.org/2019/rust-lifetime-visualizati...
- _bxg1 7y agoThis is a good subject to discuss, but the article doesn't suggest much in the way of specifics. What kinds of libraries (or library refinement) are most wanting? I agree that there's a general green-ness to the current offerings, but it isn't clear where a motivated individual might best spend their energy. In addition to speed and lifetime visualization, another big thing I'd like to see from rustc/RLS is better-localized error messaging. I have a frequent experience where an error in one place will suppress (or cause!) a completely unrelated error somewhere else, even when those bits of code don't interact at all. For example, one function that returns the wrong type may suppress the fact that a different function, in a totally different file, has a similar problem. This makes it very hard to squash errors in a methodical way. It starts to feel like whack-a-mole; traditional boundaries like function signatures don't seem to apply to compiler errors.