I'm super proud that when someone starts listing why they like #RustLang, compiler errors are always amongst the first 5 things they say. It's easy to get disheartened that there's so much left to do, and that you have to fight entropy as things get refactored and specific diagnostics regress. But a reminder from people with fresher perspectives is super helpful. Comparing #Rust to what it was a few years ago reminds me of the power of small incremental improvements over time.
It's easy to get disheartened because my view is that of what's left to do, how many open tickets are left and how many are created every week. You take on a very negative view of "the product" because you know how the sausage is made. But, with effort, it can still feel "magical" to users, as long as it keeps improving.
The best thing is that the higher you raise the bar, it attracts people who care about this, and they get used to used to it, so they demand more, pushing us to get better than we otherwise would. It's a self-reinforcing cycle.
But it is work, and one that is hard to make the time and monetary investment case for.
No-one *chooses* a tool because of something as amorphous as "good DX". But it makes a difference on whether someone keeps using it enthusiastically or dreads it.
@ekuber anecdotally, part of why I picked up Rust in the first place was seeing my friends talking about it enthusiastically
@cguille I should have said "no-one who chooses tools for others to use" :)
@ekuber let me be Arya :(
@ekuber i do actually choose tools based on DX lol
@jyn I should have said "no-one who chooses tools for others to use" :)
@ekuber Actually, when I was working on DevX we found over and over that people would claim to choose platforms and tools because of things like security, coverage, etc, but when you looked at how they actually made the decisions it was nearly always about experience.
@ekuber I have a pretty wide view of all of open source .NET and I can say: everyone has an ever-growing backlog. So don't worry if that's yours as well. The domain/area of the project doesn't seem to matter either.
@ekuber If you feel disheartened remember that there is no single compiler out there with as good diagnostics and error message as rustc :)
@epilys if all Rust accomplishes is shaming the rest of the industry into spending the effort to improve and surpass us, it will have been worth it. And I doubt that will be the only accomplishment.
@ekuber Yeah :)
Let's not forget that C++ is still taken seriously as far as PL design goes:
@ekuber
Rust is still miles ahead of any industrial programming language in it's current situation in terms of error messages. So don't feel bad about the missing part!