(1)The use of standard UTF-8 charset means the non-shortest UTF-8 form of BMP
   character is no longer accepted.
 
    -A desirable change, see http://ccc.sfbay.sun.com/4486841 for details.
 
(2)Also the supplementary characters in file names and comments are now output/encoded
   in standard 4-byte UTF-8 form. Jar tool and java.util.jar/zip package in earlier
   releases do not understand the 4-byte form.
 
    -It is the "incompatibility" (forward compatibility) that we can't avoid if we want
    to update to the latest UTF-8 standard and make the Java Jar/Zip file exchangeable
    with other standards-compliant Zip implementations. The price we have to pay.
 
    -The standard UTF-8 charset does accept pair of 3-byte surrogates when decoding, so
     we are backward-compatible
 
    -A modified-UTF-8 charset implementation can be added into our java.nio.charset
    repository should such request/escalation come in (in which case you can use the
    modified-UTF-8 and the proposed APIs to generate "old-style-zip-jar" file).