Michael Ficarra

I think the more useful project would be a schema validator, so no matter how you generate an AST, you can make sure it conforms to the spec. In order...

@RReverser I spoke to @bterlson last night about his interests, and he said he would be willing to accept general parsing pass/fail tests into test262. We should put together a...

Sure. This could be a good topic for the upcoming jQuerySF face-to-face meeting.

Remember, `SourceLocation` (which all `Node`s may have) ``` interface SourceLocation { source: string | null; start: Position; end: Position; } ``` has an optional `source`. Unfortunately, it would require you...

I don't think raw is best here. It adds a lot of burden for something that can be fixed much more simply with one of the above suggestions.

As far as codegen, `2e308` is the common "canonical" representation of Infinity for doubles. So that's not really a problem. But you should consider whether most usage will care to...

:+1: jslint is and always has been a terrible tool.

The problem with this is that support for ES3 (with respect to identifier names and reserved words) was so poor that this would be unrepresentative of any implementation. So I...

I think you're right. It seems like ```js switch(f()){} ``` will be incorrectly dropped as dead code. @aryx Are you interested in sending a PR?

You probably want to use [escope](https://github.com/estools/escope) for this.