Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: Characters losts during Import (8.1.6) -> (9.2.0.6)

Re: Characters losts during Import (8.1.6) -> (9.2.0.6)

From: Sybrand Bakker <postbus_at_sybrandb.demon.nl>
Date: Wed, 05 Jul 2006 22:25:36 +0200
Message-ID: <jq7oa2dnin7481akjjlts86am6ji9ba783@4ax.com>


On 5 Jul 2006 10:36:46 -0700, "Anurag Varma" <avoracle_at_gmail.com> wrote:

>What do you mean by sql*loader uses the same mechanism?
>Consider this:
>
>* exp done using NLS_LANG us7ascii. Oracle will record in the dmp
> file that exp was done using us7ascii. If imp then uses 8859, oracle
> will report charset conversion.
>
>as compared to:
>* Data unloaded with nls_lang as us7ascii (no charset conversion as
> same charset is being used).
> Then use sql*loader and specify nls_lang as 8859. Again, no charset
> conversion.
>
>the difference is that oracle exp will record the nls_lang setting.

I don't agree with you.
In the exp case 8 bit characters will not be stripped if database is 7 bit and NLS_LANG is 7 bit. This is likely the OPs situation. So exp will contain 8 bit data, as there is no conversion.

In the second case, as NLS_LANG has been set to 8-bit, sql*loader will perform charset conversion. The only difference between sql*loader and imp, is sql*loader doesn't report charset conversion.

In *both* scenarios there is charset conversion, whether you like it or not. However, in both cases the source data is NOT converted.

--
Sybrand Bakker, Senior Oracle DBA
Received on Wed Jul 05 2006 - 15:25:36 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US