Danny

I've never been a fan of the specialized syntax for promises in JS. Something like generators (coroutines) or do-notation are much more general and flexible. We'd do newcomers a disservice...

I really tried hard to do just that. 🤷‍♂️ The example I gave is a minimal case to reproduce this problem. Don't think the browser matters because the values are...

Yeah, that's basically the `#2` option. Certainly works, just not as automatic as it is with solutions like styled-components. Also gets a bit hairy when doing SSR. You either have...

> Postcss is doing the css-modules work, so one option is to write out or inline the json map of the css classes and then write the css out as...

> we don't want to auto import and append to the DOM b.c that would remove a consumers ability to extend or consolidate. I think many would prefer to sacrifice...

Agreed. In addition, running `--fix` with this rule enabled completely mangles your code.

Yeah, this is really a fault of the next.js `Link` API. I submitted an issue. https://github.com/zeit/next.js/issues/5533

The problem with refactoring JSX-returning functions into components is that it can break your tests if you are shallow rendering. For example, in the first case, I could write test...

Hmmm...I see your point in where you refactor out reusable components, but in the case where you just need "helper components" which will only ever be used by one component,...

Now your test is aware of the implementation details of your component. 🙁 Decide that helper component isn't needed, or split it into two, and your tests break again. BTW,...