Hi all, We discussed this change in a previous e-mail thread. Here’s a patch for your consideration: http://cr.openjdk.java.net/~tonyp/8151098/webrev.1/ I cloned the Cache class from ThreadLocalCoders and reworked it a bit. The StringDecoder and StringEncoder classes had some common fields (the Charset and the requested charset name). I moved them to a superclass (StringCoder) which made the cache easier to write (I didn’t have to create one subclass for the decoder and one for the encoder, as it is the case in ThreadLocalCoders). Feedback very welcome! Tony ----- Tony Printezis | JVM/GC Engineer / VM Team | Twitter @TonyPrintezis tprintezis@twitter.com