Character Encoding and File Encoding issue

Question

Hi,
I have a file which has a data encoded using default locale.
I start jvm in same default locale and try to red the file.
I took 2 approaches :
1. Read the file using InputStreamReader() without specifying the encoding, so that default one based on locale will be picked up.
-- This apprach worked fine.
-- I also printed system property "file.encoding" which matched with current locales encoding (on unix cooand to get this is "locale charmap").
2. In this approach, I read the file using InputStream as an array of raw bytes, and passed it to String contructor to convert bytes to String.
-- The String contained garbled data, meaning encoding failed.
I tried printing encoding used by JVM using internal class, and "file.encoding" property as well.
These 2 values do not match, there is weird difference.
For e.g. for locale ja_JP.eucjp on linux box :
byte-character uses EUC_JP_LINUX encoding
file.encoding system property is EUC-JP-LINUX
To get byte to character encoding, I used following methods (sun.io.*):
ByteToCharConverter btc = ByteToCharConverter.getDefault();
System.out.println("BTC uses " + btc.getCharacterEncoding());
Do you have any idea why is it failing ?
My understanding was, file encoding and character encoding should always be same by default.
But, because of this behaviour, I am little perplexed.

Answer 1

But there's no character encoding set for this operation:baos.write("���".getBytes());