Here is the third preview of the Unicode library developed by myself and mentored by Eric Niebler as part of the Google Summer of Code, and probably the last one as part of this program. (I will continue development and maintenance of it, however) The docs are available online at: http://blogloufoque.free.fr/unicode/doc/html/ The code, documentation, examples, tests and data parser are also available on the sandbox subversion repository, and as an archive on the vault: http://tinyurl.com/n9xqju Note you will need to set BOOST_ROOT to the root of your boost trunk checkout in order to compile the library, examples, tests, documentation or data parser. One particular issue subject to debate lately is the name of the "Pipe" concept. It has been suggested it be renamed to Transducer, Transcoder or Converter. I would appreciate community feedback on this. I'm also trying to polish documentation, so I would like to know anything that could be made better, or anything that is a bit unpractical or hard to understand. (and of course, things that are incorrect as well) You will also notice part of the reference is separate, and that some files do not lie under boost/unicode. Those are things I am pondering whether they should get included in Boost.Iterator or Boost.Range instead of Boost.Unicode. any_range (which depends on any_iterator, that I didn't write) and join_iterator certainly ought to not be in the Unicode library, but I'm unsure about the rest. There are also a few little things in the source I'd like to get included elsewhere, such as foreach_auto, common_type, static_pow and mpl/ustring, but I didn't document these at all.