CharSet Ambiguity !!!

Question

Dear All ...
I have a big confusion regarding playing with charsets in Java in general...
My problem is that I am working on arabizing the Lotus Notes Portlets, I finally reached to arabize the whole portlets (static & dynamic strings, i mean by dynamic those returned from the Domino server) but by altering the code of the java Portlets, and using a ServletOutputStream as the out object to write with on the rersponse object rather than a PrintWriter out object, the ServletOutputStream worked and the strings returned from Domino are now displayed in arabic inside the four portlets as desired, but PrintWriter didn't worked with me. Although i understood after reading in several Java Documnets concerned with Java I/O streams that the PrintWriter is more powerful than OutputStream, it can support more character sets and deals with unicode.It is written that the Reader and Writer Family of calsses are easier to internationalize because they are not dependent upon a specific character encoding. So My question is which is better -practically in terms of internationalization- PrintWriter or ServletOutputStream?
My Second qustion : Can I know the encoding of a certain retreived string (e.g str.getEncoding() or a similar method), and is it really possible to change the encoding of a string from any charset to any charset using the follwing algorithm :
String str = <arabic string to be displayed>
byte[] bytes = str.getBytes("WINDOWS-1256");
String subj = new String(bytes,"WINDOWS-1256");
out.println("Your string is :"+subj);
Is there restrictions when converting a string from a charset to another?
My third question, and the last: when using ServletOutputStream out for writing on the response object , and retreiving the translated strings from a ListResourceBundle it gives an ERROR OCCURRED! message on the rendered portlet show mode, but when retreiving using PropertyResourceBundle it worked, using a PrintWriter the opposite scenarion worked with me.
The problem with ListResourceBundle is that you cannot provide more than one bundle for the portlet, and hence you cannot provide a bilingual interface. When I tried to specify more than one resource tag under the portlet tag, it gives an error while parsing the provider.xml, and if you provide the following code:"
Locale l = pr.getLocale() ;
ResourceBundle b ;
b = pr.getPortlet().getResourceBundle(l);
It always contacts the ListResurceBundle class that you mentioned in the resource tag -which contains arabic strings in my case -regardless of the current local whether it en or ar.
Can any one give me a hand,to clarify these issues, I'll apreciate his help very much.

Answer 1

1) PrintWriters are easier to deal with. When you create your PrintWriter, you specify an OutputStream AND a character encoding. Anything written to the PrintWriter will be converted to the specified character encoding and then written to the OutputStream.
When reading a file, you can do the same - create a Reader using a specified character encoding. The file/stream will be read using the specified character encoding and converted into Java characters (char data type). All chars & Strings in Java are in Unicode (UCS-2) so once you have read data into a String, using a Reader, all your data is in a consistent format.
Use the Writer to write that data out to another format. Using Readers & Writers means you only have to worry about character encoding when you read & write data. Everything in between is just Strings.
2) Answered above. Strings are always in Unicode (UCS-2). What matters is how you create the String & how you write it. If in doubt, be explicit when reading/writing
3) You cannot change the encoding of a String - Your code will do the following:
a) convert a UCS-2 String to bytes in Windows-1256
b)create a new String (in UCS-2) from a byte array encoded in Windows-1256
4) When using ResourceBundles, all your translations should be part of the same bundle. You do not have multiple bundles.
Look at the HelloWorld sample in the JPDK - the one that has a "resource" directory. It has one resource bundle & multiple translations of that bundle - the translations are accessed based on the locale you request.