Default Character sets on Windows

When you use the DBCA to either create a DB or create the scripts to create a DB I presume it picks something up from the OS to help it determine which character set to use when creating the database.

My question is: what does it use?

I have 2 DBs - one 'DB1' on a Windows NT4 server, one 'DB2' on a Windows 2003 server. The character set on DB1 is WE8ISO8859P1, the character set on DB2 is WE8MSWIN1252 and AL16UTF16 NCHAR.

I'm a novice at this type of stuff so would appreciate some pointers.
Would an export from DB1 imported into DB2 give me any real problems? The import works okay but I get a line in the log file saying "possible ncharset conversion".

I think DB1 is probably Oracle 8i db, whereas DB2 might be Oracle 9i db

Exactly this is because of different versions of oracle and also the import works because WE8MSWIN1252 is a superset of WE8ISO8859P1.

Correct about the versions chaps!
Can you elaborate?

Should the import/export function ok? Or will there be possible character conversion problems?

Oracle takes default characterset as WE8ISO8859P1 for 8i and WE8MSWIN1252 for 9i during their database creation.Since both are supersets u willn't have any problem with the data even though it mentions as possible conversion at the time of import.

We have a Trademark symbol ™ in our 8i database. I did an export and imported it into a 9i database. It appears in the 9i database as ¿. I know ¿ is the default character when Oracle can't transalte a character.Hence my concern.

Anyone have any ideas or comments as to why this happened? I know ™ isn't a transportable character but if 8i and 9i charsets are compatible why can the exp/imp not translate the char?

Quote:

---

Originally posted by akhadar
Oracle takes default characterset as WE8ISO8859P1 for 8i and WE8MSWIN1252 for 9i during their database creation.

---

Not quite. Not even close!

Oracle will check your OS regional settings and decide which character set is appropriate for you by default.

On a Windows server any idea what regional settings it checks?

control panel -> regional settings -> (location and/or language settings)

Both have regional settings of English (UK) in the Windows OS. Which brings me back to my post which may have been missed:

Quote:

---

Originally posted by JMac
We have a Trademark symbol ™ in our 8i database. I did an export and imported it into a 9i database. It appears in the 9i database as ¿. I know ¿ is the default character when Oracle can't transalte a character.Hence my concern.

Anyone have any ideas or comments as to why this happened? I know ™ isn't a transportable character but if 8i and 9i charsets are compatible why can the exp/imp not translate the char?

---

I'm just confused why 1 symbol in the 8i DB wasn't imported correctly into the 9i DB when both have apparently compatible character sets?

Quote:

---

Originally posted by akhadar
Since both are supersets . . .

---

:confused: :rolleyes: :confused: :rolleyes: Two sets can't be supersets of each other unless they are identical!

Quote:

---

Originally posted by JMac
. . . both have apparently compatible character sets?

---

I suspect not - my guess is that neither is a superset of the other. If the WIN charset was a superset of the ISO one, Oracle wouldn't give the warning.

Does any one have a good reference showing the actual code tables?

have you checked the client side settings from where you are accessing this character

Open up regedit->HKEY_LOCAL_MACHINE->SOFTWARE->ORACLE->HOME0 and check the value for NLS_LANG.

sorry for the trash which i posted early.
After lettle bit search i found some useful information which deals with WI8MSWIN1252 and WE8ISO8859P1 character sets on two databases and what u can do to correct the data which is shows as ? after import.
you can run csscan (oracle util) to scan the datafiles for characters which may not map cleanly from 1252 to ISO-1. To fix those, you may need to run a replace() on the data and then export it.

Try setting the NLS_LANG=american_america.WE8ISO8859P1 on windows before exporting. That way the export file will contain WE8ISO8859P1 data.
Isolate the � using substr() and get it's decimal value using dump(). It seems to be Ox93 (using windows Character Map for Windows:Western). The right hand one seems to be 0x94. The regular " is Ox22.

update my_tab set mycol = replace(mycol, chr(123), chr(234))
where mycol != replace(mycol, chr(123), chr(234));

Also check this link
http://metalink.oracle.com/metalink/...&p_id=137127.1