> As you can imagine, Ruby code intelligence (type inference, etc.) is super tough, and there are maybe 5-10 people in the world who could build it right now.

Is Yin Wang in the list?

Do you know what happened to RubySonar?

I remember reading something about it, but now only search engines retain traces of it.

Here is his post about why he closed source the PySornar and RubySonar: http://www.yinwang.org/blog-cn/2017/04/18/close-source

Thank you.

Too bad it's in Chinese, but Google Translate got me the gist. An unfortunate story.

Not sure why he didn't settle on GPL or AGPL, though.

@whitequark and @backus are the first that come to mind. But anybody on the ruby core team, working on parser or even rubcocop could do it.

Disclaimer: I contributed to rubocop.

What!? There are libraries available that do static type inference on Ruby. Although the language is rather dynamic, it's in no way an intractable problem, or even one that requires a PhD in Ruby Bullshit.

Here are some libraries that could contain that one-in-a-billion knowledge:

- RDL: type annotation, but also has a static type checker. https://github.com/plum-umd/rdl#static-type-checking

- Typed: gradual typing system, with type inference https://github.com/antoniogarrote/typed.rb

- Jetbrains has something, too: https://github.com/JetBrains/ruby-type-inference

Thanks. I am on your side and have been a Ruby fan since 2005. I'm trying to throw money at building better Ruby tooling. :)

I'm familiar with those projects, but when I last tried them, they did not work well enough with some additional constraints that the nature of our product imposes (chiefly requirements for error-tolerance and cross-project resolution), and we weren't able to find people to help us. I will try them again and see if we can sponsor someone now to help integrate when with Sourcegraph, since it looks like quite a bit of progress has been made in the last few months.