[PATCH] Identify language and country based on LC_MESSAGES rather than LC_CTYPE

Hi, The attached patch changes the code that currently parses LC_CTYPE to set encoding, encoding variants, language and country to use LC_CTYPE just to set the encoding, and instead uses LC_MESSAGES to set country and language. The reason for this change si that in a system where LC_MESSAGES is set to English, you'd expect Java to answer in English, rather than Italian by default (this is true for both JDK's commands and for Java-based interfaces). The patch is probably rough, so I'd be happy to hear comments on what to improve of it. -- Diego "Flameeyes" Pettenò http://farragut.flameeyes.is-a-geek.org/