From WE8ISO88559P1 to AL32UTF8

Question

I have 10.2.0.3 databases on UNIX
my source database has characterset of WE8ISO88559P1, My target database has characterset of AL32UTF8.
Do you think there is going to be some database loss?
I mean any character in WE8ISO88559P1 can find a match in AL32UTF8.
Thanks!

Answer 1

1) Unless you set NLS_LENGTH_SEMANTICS to CHAR before creating your table, creating a table that has a VARCHAR2(10) column allocates 10 bytes of storage, not 10 characters of storage. In the UTF-8 character set, one character may require up to three bytes of storage space, so a VARCHAR2(10) may only hold 3 UTF-8 characters. In general, you have three options
- Set NLS_LENGTH_SEMANTICS to CHAR before creating all your tables, thus creating all columns in terms of characters.
- Increase the size of all CHAR/ VARCHAR2 columns by a factor of 3
- Explicitly declare all columns using character-length semantics (i.e. VARCHAR2( 10 CHAR)
2) The character set conversion that will happen will depend on the client's NLS_LANG setting. You need to set the NLS_LANG to identify the character set of the data you are sending and to identify the character set of the data you are attempting to retrieve, otherwise the conversion process will not work as you'd expect (i.e. accents will be dropped and/or characters will be deemed unconvertable and rendered as question marks or upside down question marks).
Justin