Experts: ****Urgent**** help needed.

Question

We need to give a demo to a customer tomorrow.
We need to store data in the Oracle 10g DB in Persian.
Our development DB which we inserted data in English has a WE8ISO8859P1 character set. There is no data in Persian in this DB.
We created a new 10 DB with char set AL32UTF8.
Now we need to import data (From WE8ISO8859P1 dev DB which has data only in English) to AL32UTF8 DB.
We will start entering in Persian only in the AL32UTF8 DB.
All our columns in tables are just default VARCHAR2 columns.
Will we have to change the columns of our tables in AL32UTF8 DB to NVARCHAR2?
Will there be any problems exp/imp???
Will there be any problems in entering data in Persian in AL32UTF8 DB?
Will a column in WE8ISO8859P1 which is VARCHAR2(10), allow to enter 10 Persian characters in the AL32UTF8 DB?

Answer 1

Thanks Experts for the answers.
I finally got hold of a AL32UTF8 DB, changed my machine's language to Farsi and started inserting records and discovered these things which you also have mentioned.
(1.) One Farsi character will take up 2 bytes.
(2.) In AL32UTF8 if you declare a field as VARCHAR2(1 CHAR) it will allocate 4 bytes for that. If you just say VARCHAR2(1) only one byte will be allocated.
(3.) You can declare a field as (maximum) VARCHAR2(4000 CHAR). Then maximum bytes you can store is 4000. i.e. only 2000 Farsi characters. If you enter 2001 Farsi characters it will give "value too large for column" error.
(4.) So, if you have a WE8ISO8859P1, and have a table which as a VARCHAR2(10) field. You use a DDL script from WE8ISO8859P1 and run it just as it is in AL32UTF8. Now, if you EXPORT from WE8ISO8859P1 and IMPORT to AL32UTF8 there wont be any problem, since WE8ISO8859P1 is a single byte character set.
(5.) Here is the problem. There is a NAME VARCHAR2(100) field in AL32UTF8. Now in AL32UTF8 you will be only be able to enter 50 Farsi characters into that. Solutions:
(5.1) In AL32UTF8 recreate the column as NAME VARCHAR2(100 CHAR). I did it this way for the demo.
(5.2) When creating the AL32UTF8 DB create it with NLS_LENGTH_SEMANTICS set to CHAR. So you can just use DDL extracted from WE8ISO8859P1 (i.e. NAME VARCHAR2(100)) in the AL32UTF8, since in AL32UTF8 100 will go as 100 CHAR. This way you don't have to modify the DDL scripts. Only drawback is for VARCHAR2 fields which need more than 2000 Fasi characters to be input, you will have to make those VARCHAR2 fields to CLOB. i.e. any field which is VARCHAR2(2000) or greater in WE8ISO8859P1 will have to be changed in AL32UTF8 to CLOB.
I find this method (5.2) the better method. But I read somewhere that it is NOT ADVISIBLE TO set the NLS_LENGTH_SEMANTICS to CHAR at DB creation time.
Is this true? If so why??