It should be no real problem to have Java create UTF-8 encoded files/streams (both JSON and XML), and JSON can read UTF-8 data just fine... (Similarly for XML)

Agreed. As long as no UTF-16 chars are used, which is really easy to do in java since strings are all UTF-16, it's all gravy.

Java strings are UTF-16

A good XML writer that prevents you from going outside of the declared format will protect you from future mistakes. Down side is the extra effort in using some XML api over others. JSON is usually gravy in all languages.

Gravy.. mmmm...

Java strings are UTF-16

The way strings are stored internally doesn't really matter.

While Perl stores unicode strings internally as UTF-8 (or something very close to it), it can encode those strings to many other encodings for output.  The same holds for Java: while it stores strings internally as UTF-16, there's no problem creating UTF-8 output, for example.

Writer utf8out = new BufferedWriter(
                     new OutputStreamWriter(
                         new FileOutputStream("outfile"), "UTF-8"
                     )
                 ); 
utf8out.write("some unicode data");
[download]