Re: convert CL8WIN1251 to UTF8

"JC" <jeffc_at_sympatico.ca> wrote in message news:Q6C39.1966$Xf7.490916_at_news20.bellglobal.com...
> koert54 wrote:
>
> > For once I support Sybrand's statement :-) (no offense Sy-dude)
> > Anyone claiming things like 'but on page x on this book it states ...'
> > without even trying it out makes
> > me smile :-)
>
> What doesn't make me smile is when two seperate publications by Oracle
> contridict one another.

Well, it's been known for Oracle doco. to be wrong, big time. But usually it's consistently wrong! Which two publications are you comparing, though? You mentioned an "Oracle 8i Certified Professional book" as one of them: that wouldn't be an Oracle Press book would it? 'Cause if it is, Oracle Press are not a part of Oracle Corporation, and therefore Oracle itself is not making contradictory statements.

Anyway: that aside, it has been possible to change charactersets for a long time, but the restrictions on doing so are pretty major. There must be a code-point mapping one-for-one between the old and new character sets (ie, the new one must be a superset of the old one, so Japanese to Russian is out). Otherwise it doesn't work. Converting the character set also means it's out of commission to the users for the duration. There is a risk of data truncation (char(5) means 5 BYTES, not characters, so converting to a double-byte character set means words like 'Howard' would get converted to 'How', which is not pleasant).

All in all, it's such a tricky and thankless task that the advice from Oracle to get the decision right in the first place is not out of order or incorrect.

Incidentally, you can use the csscan utility (character set scanner) to work out the problems a conversion would bring *before* actually attempting the conversion. That would have shown our original poster that his proposed conversion wouldn't have been possible, too.

Regards
HJR
>All the study material I get from Oracle (and
> outside Oracle) states it's important to set NLS stuff when you create the

> database because it can't be changed after.
>
> If also mentioned that I could be wrong because I have seen other errors
in
> Oracles OCP exam study material. I was interested to find out about it.
> Im not going to go through the hassle of creating a database then changing
> the NLS paramaters before I post a message in the newsgroup. You could
> politely reply by saying "Yes, Oracle got that wrong in that publication".
> Thats what I was expecting to hear instead of ego gloating like the above.
Received on Tue Aug 06 2002 - 04:33:52 CDT