Release Notes.txt

This library provides UTF-7 and Modified UTF-7 Charsets for Java.

Sun Java 6 is recommended.

Sun JRE versions before Java 6 have two bugs which make this an imperfect 
implementation:
- due to bug #6221056 flush is not always called, and the last few bytes of 
  an encoded sequence may be missing if CharsetEncoder.encode(ByteBuffer) 
  is used.
  A hack is used to get around this: OVERFLOW is returned from the 
  encodeLoop if the JVM detected is a 1.4 or 1.5 version, and bytes are 
  waiting to be flushed, and the amount of bytes allocated in the output 
  buffer does not indicate String.getBytes (or similar) was used; this
  last method tries to always allocate sufficient bytes, and fails if
  OVERFLOW is returned. Unfortunately a clean solution was not available, 
  since the buggy method is declared final.
  See Also: http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=6221056
  Please note this is not a problem if the last of the sequence is a 
  directly encoded character in UTF-7, e.g. an alphanumeric character, 
  space, or CR/LF.
- additionally, you should take care before calling flush that sufficient 
  bytes are available in the output buffer. Reinvoking the flush method of a
  Charset(En|De)coder throws an IllegalStateException, even if it returned 
  OVERFLOW the first time.
  
gij (GNU libgcj) version 4.1.1 seems to suffer a similar bug, due to which 
the flush method is not called. The hack I described earlier does not work 
here.