Re: Export-Import across character sets

Kenneth C Stahl <BlueSax_at_Unforgettable.com> writes:

> Sybrand Bakker wrote:
> >
> > Especially for US citizens:
> > Please note WE8ISO8859P1 is equivalent with the Latin-1 alphabet and has
> > definitely not to do with Iceland only.
> > On NT WE8ISO8859P1 is already the default and rightly so as it covers the
> > major European languages.
> > In an internationalizing world, with many international firms with
> > international data in their databases, creating a database in US7ASCII is
> > very undesirable, not to say more, not to say it is typically US-centric.
> >
> > Best regards,
> >
> > --
> > Sybrand Bakker, Oracle DBA
> > Kenneth C Stahl <BlueSax_at_Unforgettable.com> wrote in message
> > news:384FCC02.8E51C83D_at_Unforgettable.com...
> > > For some reason that totally escapes me, a data base that I work with was
> > > created with the WE8ISO8859P1 character set instead of US7ASCII. Does
> > > anyone know if there would be adverse effects if an export were taken from
> > > that database and then imported into a database with the US7ASCII
> > character
> > > set? We are no where near Iceland so nothing has been inserted into the
> > > database which would use Icelandic characters.
>
> Then the followup question would be: Does Oracle have any definite guidance
> on this subject? I want to establish a standard, and if WE8ISO8859P1 is
> better then US7ASCII then I'm willing to do this, but I want to do it for
> the right reasons rather than on a weak reason that the US7ASCII is
> US-centric.

The risk is that west european applications, that have NLS_LANG set to a 8 bits characterset, lose the 8th bit of their characters. This means character over ascii code 127 (eg יטאהצ ...) will be truncated, and if they query the table, they will get different results back.

The adverse effect is actually that if the database is in US7ASCII, and the client is also in US7ASCII, you can transparently insert characters greater than ascii code 127, and the characters are NOT converted because the client and the database's characterset are the same. However a client with NLS_LANG set to WE8ISO8859P1 will see the characters differently.

As a summary I always create a database in a 8 bit characterset, especially WE8ISO8859P1, because it is a strict superset of US7ASCII.

>
> By the way, there are a series of scripts in the
> $ORACLE_HOME/rdbms/install/rdbms directory the file
> rdbms.vrf lists WEISO8859P1 as the Icelandic character set.

--
The fundamental aim of King Crimson is to organize anarchy, to utilize

   the latent power of chaos and to allow the varying influences to

	       interact and find their own equilibrium.
                                        -Robert Fripp