The encoding converter API (src/share/classes/sun/io/CharToByteConverter.java) specifies the following behavior when unknown characters are encountered in the input: If substitution mode is enabled, the unknown character is mapped to the substitution bytes for the converter. If substitution mode is disabled, an UnknownCharacterException is thrown.

None of the motif encoding converters implement this behavior as specified. Most map unknown characters to random, implementation-dependent bytes. The Dingbats converter throws the exception independent of the substitution mode.

This bug makes the current implementation of multi-fonts completely unusable - text that needs multiple fonts is at most partially displayed. See bug 4072782 for details.