project-error-handling VS goformat

This actually is an example of where the compiler errors could (or should have) maybe provided more help or even the potential solution, it might be worth submitting this to the error handling group.

If anyone's interested in helping to shape the future of Rust's built-in error-handling story, there's an error handling project group that's been doing great work recently, e.g. the major effort to move the Error trait into libcore ( https://github.com/rust-lang/project-error-handling/issues/3 ) and stabilizing std::backtrace. You can follow along or get involved via the #project-error-handling channel on the Rust zulip: https://rust-lang.zulipchat.com/

Getting it into alloc would enable usage in a LOT more contexts, like WASM and kernel code. Does this need a distinct tracking issue outside the ticket for moving it to core or would that just add more administrata?

without trolling https://github.com/rust-lang/project-error-handling exist and is far from having strong conclusion and anyway I will always favor enum Error anyway however I like the idea to have a opaque box in the enum for "this is a opaque error you can't deal with as a user of my api"

IIRC, the Error Handling Project Group is aware of these ideas. If this kind of thing interests you and you want to contribute, you should look into getting involved with that group.

> you should have the underlying message of the std::io::Error

This is a point of debate[1] among the error-handling working group.

[1]: https://github.com/rust-lang/project-error-handling/issues/4...

Focusing on good error messages has permeated throughout the community. There's even the Error Handling Project Group if you weren't convinced how committed the language designers are to getting this right. There are a number of techniques we can use to make our errors more informative. Along the way, we will discuss the crates that can help.

The Error Handling Working Group is looking at potential breaking changes for embedded users. Maybe you could work within that group?

Likewise, gofmt implementing what you argue for resulted in the creation of goformat. There's a limit to how much you can force people on these things and, more importantly, there are formatting decisions which are more than mere bikeshedding in the eyes of the programmers to the point where they consider it more productive to maintain the formatting by hand if that's what it takes.

I'm the guy who would only run rustfmt once every week or so, when my codebase was in a clean state where I could use git gui to cherry-pick the changes that were in line with my stubborn insistence on my own style and revert the rest. I'm also the guy who would have considered writing goformat if someone else didn't.

Enough people disagree with that for goformat to exist.

Same. I used to insist on cargo +nightly rustfmt and a massive stable of "I don't have a portrait-oriented monitor" rustfmt.toml tweaks which I'd only apply when I have a clean git gui on hand to cherry-pick away unwanted changes, but I've mellowed out and the rustfmt handling of things like assert! has evolved so, now, I just put use_small_heuristics = "Max" in my rustfmt.toml as an analogue to the people who choose goformat over gofmt.

anyways, regardless of what I think, perhaps this library could help? Or at least be a good starting point to build your own: https://github.com/mbenkmann/goformat

You'd prefer that people like me follow the road the Go ecosystem did and write goformat as a replacement for gofmt or just continue to hand-format everything?

Because Go syntax is ridiculously simple, there's not much room for opinion. And even considering that, there is already an alternative gofmt with custom options.