JRuby has had a limitation up through JRuby 9.1.x where some indentifiers which do not properly map to a Java charset are not possible beacuse internally we store all identifiers as Java Strings.

The solution to this is partially to move as much internally to using ByteList as we can. For things where this would not work out well we will still use java.lang.String but we will store those strings as an iso_8859_1 raw piece of data. Anything which accesses those String will need also to have their encoding and if they don't we will assume they are utf-8 data. This largely will allow things we could not do in the past to start working but things from the past which did work will still work since pretty much everything is utf-8.

• [JRuby creates symbols with US-ASCII encoding but non-ASCII bytes #4828 ]
• [ RuntimeError with compiled multi-byte variable RuntimeError with compiled multi-byte variable #4660 ]
• [ define_method using symbols string syntax works incorrectly define_method using symbols string syntax works incorrectly #3880 ]
• [ encoding of utf8 symbols (eg. :λ) encoding of utf8 symbols (eg. :λ)  #3719 ]
• [ Identifier-keyed tables must always use raw or always use encoded identifiers Identifier-keyed tables must always use raw or always use encoded identifiers #3697 ]
• [ Multibyte identifiers not marshaled correctly Multibyte identifiers not marshaled correctly #3688 ]
• [ method does not return valid symbol when Non-ASCII method __method__ does not return valid symbol when Non-ASCII method #914 ]
• [The & operator doesn't work with quoted symbols The & operator doesn't work with quoted symbols #5066]