RegExp character classes are incorrect

[ChALkeR]

/\w/.test('\u212A')
/\w/u.test('\u212A')
/\w/i.test('\u212A')

true under nova, false under v8 (and spec)

Note that /\w/iu.test('\u212A') is true under v8 and per spec.

[aapoalas]

Right, yeah; this relates to the RegExp engine issue in general (and I added a tag for that to group them together). The regex engine is set to run with the unicode flag always, as it is closer to conformant in that setting.