Re: National Language Support (NLS) questions.

On Mon, 29 Sep 1997 18:18:40 -0400, Yuri Korostyshevski <ykoros_at_dowjones.net> wrote:

>
>I am interested in the case where there is a need for conversion
>between character sets with little or none common characters.
>(e.g. Bulgarian and Chinese)

There are basically two types of charactersets. One, Single byte and the other, Multibyte.

If the client and the server have both different single byte charactersets, Oracle uses lookup tables to perform the conversion. This lookup table is in the software and I do not believe one can program around it.

If a Multibyte characterset in involved, Oracle uses routines to perform the conversion.

The qs becomes, how will Oracle convert Bulgarian to Chinese? As Bulgarian is an East European language, am assuming it will have some mapping to the 'English' characters. As ascii is part of all charactersets, most Bulgarian characters will be transformed to ascii chars (which is part of the Chinese charset).

In other words, Bulgarian is not being transformed to Chinese, but rather the most appropiate chars in the Chinese charset, which includes ascii.

>The questions I have are:
>Is it possible to define replacement characters between ANY two
>character sets?

No.

>Where and how do I define the replacement characters?
NA

>Where and when the conversion takes place, and exactly which
>parameters govern it?

The conversion takes place in one of the SqlNet layers. It is all coded in the Oracle software and works tranparently.

>Any help, especially from those who have created multi-lingual
>database applications, would be much appreciated.
Use the UNICODE characterset. Oracle uses a multibyte variation of the UNICODE characterset. As UNICODE has all the characters, you should be able to create your multilingual apps.

-sm Received on Thu Oct 02 1997 - 00:00:00 CDT