Inserting and retrieving data from a al32UTF8 database USING SQL Developer

Question

hi guys,
Before i post my questions , i think its better for me to provide you guys with my understandings first so that it easier to understand where/if i have gone wrong..
I am using Window XP and Oracle 10g
Non-unicode client - a client program that need to use the OS code page for mapping of the retrieved unicode data from the database as well as the support of displaying/inserting the characters from that code page to the database.
E.G sqlplusw.exe
Therefore, when using a non-unicode client
1) we have to set the OS code page (Control panel - regional and language setting - advance - language for non unicode program ) to the code page that contain the characters we are going to display/insert.
2) we will also have to set the NLS_LANG characterset to the character set of the code page we are going to insert so that when we do a insert (for e.g in thai ) , oracle will know, and auto conversion to UNICODE can take place. This is also true when we retrieve unicode data from the database so that conversion to the correct character set can take place.
INSERTING
THAI ---> conversion ----> UNICODE
RETRIEVING
THAI <---- conversion <---- UNICODE
I hope my basic understanding is correct up till this point.
Unicode client - a client program that supports the displaying/inserting of unicode characters without the need of setting the OS code page (Control panel - regional and language setting - advance - language for non unicode program )
E.G isqlplus http or SQL developer
However,
1) There is still a need to set the NLS_LANG so that correct conversion can take place between the client and the database.
For e.g, when retrieving if we set the NLS_LANG character set to ZHS16GBK (chinese) and the data store in unicode in database is E.G (THAI) , then the conversion would be wrong .
Since it is a unicode supported client, then the NLS_LANG character set should be set to UNICODE as well.
Here come my questions
*Important - please help if you are busy and have no other time to answer the rest of the questions
*Q1) If i were to use a unicode client, what should i set my NLS_LANG character set to ?
AMERICAN_AMERICA.UTF8 ?
*Q2) Where do i set the NLS_LANG character set information in SQL Developer, i know there is a metalink for setting NLS_LANG using isqlplus but i cant seems to google any result for SQL developer.
Q3) Is my basic understanding right until this point ? If not, please explain in a more generalised term as i am really not familiar with character sets, code page, unicode , glyphs and fonts..
Q4) If a unicode client does not need to refer to the OS code page (set in regional and language) , is there a UNICODE code page for the client to refer to , or is there any Window API available ?
Q5)
There is still a need to set the NLS_LANG so that correct conversion can take place between the client and the >>database.
For e.g, when retrieving if we set the NLS_LANG character set to ZHS16GBK (chinese) and the data store in >>unicode in database is E.G (THAI) , then the conversion would be wrong .am i right on this point for UNICODE supported client ?
Thanks for spending time to read my questions and i hope to hear advices from you guys soon.
Million thanks again for sharing.
Best Regards,
Noob but willing to learn

Answer 1

The requirement to always set NLS_LANG is not true for JDBC, which ignores NLS_LANG altogether. Java programs fetch text data into String variables, which use Unicode UTF-16 by design. JDBC sets character set conversion so that data is converted between UTF-16 and the database or national character set.
The requirement to set NLS_LANG is not generally true for OCI, either. The first call in an OCI problem can be OCIEnvNlsCreate(). This call has two parameters that allow the caller to define the character set to use for VARCHAR2/CHAR/LONG/CLOB/statement text and the character set to use for NVARCHAR2/NCHAR/NCLOB. Only if these character sets are specified as 0, NLS_LANG character set is used. Also, OCI programs can specify different character sets for each bind or define variable (i.e. input/output buffer). Note: OCI programs always use NLS_LANG to initialize the language and territory settings for the client program and the database session. Only the character set can be specified is OCIEnvNlsCreate().
OCIEnvNlsCreate() can specify the client character set as UTF-16 (in platform endianess). This is not possible with NLS_LANG.
Various interfaces building on OCI, such as Oracle ODBC and ODP .NET, explicitly initialize OCI with Unicode character set, and thus ignore the NLS_LANG character set as well.
Thnx,
Sergiusz