Character Sets.

Currently there are some limitations on how Mulgara handles character encodings.

* JSON output is not utf-8 (I think it's Latin-1) (#200; was #197)
* The <?xml ...?> header in sparql xml results does not specify an encoding (#198)
* The "Accept-Charset" header is not honored; e.g. you can't use HTTP content negotiation to ask for UTF-16 or KOI-8, etc.  (#201)
* SPARQL queries with non-ascii (non-latin-1?) strings will not work; you can load such data, and retrieve it if it matches a variable, but you can't match against it explicitly.  In other words, if your graph contains e.g.  foo:bâr, then "SELECT ?obj WHERE { <foo:bâr> ?p ?obj } will not match.  (#202)