Is there a possibility to make Tern.js great again?

Over these years, TypeScript has been a huge success, and much efforts on attempts of analyzing JavaScript have gone to building a pure TS codebase instead. For example, YouCompleteMe switched to TSServer because of the lack of continuous development of Tern.

This solution has some drawbacks:

  1. Implicit any. If you don’t declare a type, it will be assumed as any type. This is completely normal in TS, but we never annotate types in vanilla JS. On the other hand, Tern is built to deal with this, and can infer the type under most cases.
  2. Performance penalty. Even querying completions in TS is a type check to the source file using the TS engine, hence the big overhead. Some build systems even treat stripping the type annotation as an optimization on bundling. We surely need a tool for this job, nothing more, nothing less.

To me, Tern is a unique piece for JavaScript. I think most developers (that know of Tern) are still using Tern in their pure JS codebase, using it as a small assistant providing fast autocompletions, jumping between sources, but it is falling short on understanding the latest revisions of ECMAScript. The most similar software to Tern that I know of is Jedi, which is an actively-maintained Python autocompletion package. As JavaScript matures, Tern can leverage modern JS constructions to provide more accurate suggestions. I wonder if anybody has an interest on reviving Tern to be the Jedi in JavaScript?