Re: Need Help: Question on Export/Import

From: Michel Cadot <micadot{at}altern{dot}org>
Date: Wed, 4 Feb 2004 18:10:46 +0100
Message-ID: <40212711$0$18986$636a55ce_at_news.free.fr>


"M" <someone_at_microsoft.com> a écrit dans le message de news:bvr801$7b6$1_at_news.hgc.com.hk...
> It seems weird that the data may contain duplicate primary key values
> because
> it was just exported from an Oracle DB with same version! The data was okay
> in the original database.
>
> I think the problem may be caused by the import process, which, somehow
> convert the data to fit the characterset of the new DB. I found that there
> was
> a line saying:
> Export done in WE8ISO8859P1 character set and ZHT16BIG5 NCHAR character
> set
> server uses ZHT16BIG5 character set (possible charset conversion)
>
> and, the data imported were simply converted to strange characters like
> "??", "???",
> which caused the Primary key validation error. Then the question is,
> why there was charset conversion when the 2 database are the same?
>
> Any clue?
>
> Thanks
>
<snip>

All depends on the NLS_LANG parameters when you do export/import. For instance, for import:
import translates the export file to the character set point by NLS_LANG while reading the file and then translates it from this character set to the one of the database.
[Quoted] I think, the '?' you get come from the first translation. Your client NLS_LANG [Quoted] don't have a transcription for some characters in your export file and change [Quoted] them to a "standard" '?'.

You must set NLS_LANG with a superset of the two character sets you use.

Regards
Michel Cadot Received on Wed Feb 04 2004 - 18:10:46 CET

Original text of this message