Re: Character set is wrong after exp/imp operation (?'s show as #'s)

"Sybrand Bakker" <postbus_at_sybrandb.demon.nl> wrote in message news:<u24lka47dgtk76_at_corp.supernews.com>...
> "David Goodall" <davidg_at_bluewave.com> wrote in message
> news:2a9f8213.0112201015.693e70d8_at_posting.google.com...
> > Hi,
> >
> > I am trying to export an Oracle database from a Linux box to a WinNT
> > box (only a temporary measure ;-) by the following method:
> >
> > ./exp system/password full=Y file=filename.dmp
> >
> > imp userid=system/password_at_dbname full=y file=filename.dmp
> >
> > The import is into a newly created database, with the characterset
> > WE8ISO8859P1.
> >
> > Everything works fine except that all instances of "£" are replaced
> > with "#"
> >
> > I then tried again with a fresh database, using the WE8DEC
> > characterset, but got the same result.
> >
> > Any help would be much appreciated.
> >
> > Cheers,
> > David.
>
> You need to set NLS_LANG on both sides as
> "<your language>_<your territory>.<your characterset>"
> of course the source database must have the correct characterset.
> If you look at exp's log output you should notice characterset conversion
> occurs when not setting NLS_LANG
>
> Hth

Thanks for that, you are absolutely correct. Having now set the NLS_LANG properly on both sides, a glance at the exp log shows the following (as you predicted):

Export done in WE8ISO8859P1 character set and US7ASCII NCHAR character set
server uses US7ASCII character set (possible charset conversion)

Which, I think, shows that the source database wasn't set up with the correct characterset? Is there anything I can do to change this?

Thanks for your time, and for putting up with a newbie... David. Received on Fri Dec 21 2001 - 05:20:35 CST