mruby does not parse escape characters in a regular expression
literal (ref #2007). It is parsed by the external mrbgem that
provides Regexp class, if any. These tests should not be in
mruby core but in the mrbgem.
This change adds the \u notation for double quoted strings and regular
expressions. It does not implement the \u notation for character literals.
Both the \uNNNN and \u{NNNN} notations are supported.
\uNNNN is implemented by emitting equivalent UTF-8; that is, "\u4000" is
equivalent to "\xE4\x80\x80".
Unlike CRuby, the \u{NNNN} notation allows only one character per pair of
braces; I see no way to lift this restriction without remodeling the parser.
Nobuyoshi Nakada