The hudson.plugins.check​style.util.JavaPacka​geDetector.detectPac​kageName()
hard-codes UTF-8 encoding when it creates the LineIterator object.

This should be amended to at least use the platform default encoding
(java.nio.charset.Charset.defaultCharset().name()). The ideal solution would be
for the user to be able to specify the encoding on the Hudson Build
configuration page.

I came across this when I ran Hudson with version 1.5 of the IBM JDK on Linux
(defect 1827). This throws an exception trying to decode UTF-8, even though the
SUN JDKs work fine. I have verified that there are no invalid UTF-8 code
sequences in any of our java files.