(Related to UNDERTOW-503 and, I assume, UNDERTOW-523)

To reiterate, XHR2 used with FormData is encoded in utf-8 and doesn't specify charset.

Currently I'm observing this:

1. since no charset is specified the form is parsed in iso-8859-1
2. servletRequest.getCharacterEncoding() returns null
   • and we convert it to utf-8
3. servletResponse is encoded in iso-8859-1
   • causing the utf-8 strings to be badly encoded

Setting default-encoding in jboss-web.xml creates additional problems:

1. form is parsed in default-encoding
2. servletRequest.getCharacterEncoding() returns null
   • and now we can't tell what the encoding actually is

Even if I could get encoding correctly the response will still be badly encoded since the initial charset isn't set - the response header is Content-Type: text/xml;charset=null. (The header charset is initially set correctly to the default-encoding but it's later overwritten, charsetSet is true but charset is null)