hikari_no_yume

We could just do what Node does, yes, and export everything with an object. On the other hand, I don't like having to look at the bottom of the file...

Right. If you implicitly export everything, then it's way too easy to accidentally expose something.

In order to do this, we need an easy-to-optimise AST. While the opcodes _sort of_ work as an AST, they aren't really. Currently we have two stages with a certain...

A rather important note: strings should be made up of true Unicode characters (codepoints), not UTF-16 code units. That means a JS string of length four containing a surrogate pair...

Another thing: it should have Unicode escape sequences with the `\u{xxxx...}` syntax I proposed for PHP: http://wiki.php.net/rfc/unicode_escape

Some sort of string interpolation would be great. I'm tempted to just have `printf` and be done with it. Perhaps `sprintf` renamed to `format`.

OK, the basics of this are done now: https://github.com/TazeTSchnitzel/Firth/commit/0a249c27593a446ea268854c00ff0e5434a2d681 All that's needed are Unicode escapes and some manipulation functions. Also, it should probably error if given invalid UTF-16 input (mismatched...

We could avoid doing any unit tests for now and only have high-level tests written in Firth itself, perhaps. Though unit tests are valuable. I'd like to get Firth self-hosting...

OK, we now have `gteq`, `lt`, `lteq`: https://github.com/TazeTSchnitzel/Firth/commit/ca66c10cc1a6e315b355a9ed84739379605b93ba However, `eq` and `neq` aren't yet implemented because they're more complicated. I thought I could just do `(a === b || (a.type...

Strings are a prerequisite: https://github.com/TazeTSchnitzel/Firth/issues/1