Re: OCTET_LENGTH is wrong

Peter Eisentraut wrote:
>
> Tatsuo Ishii writes:
>
> > > I don't think so. The sort order is independent of the character
> > > encoding, and vice versa. It must be, because
> >
> > This seems different from SQL's CREATE COLLATION syntax.
> > >From SQL99's CREATE COLLATION definition:
> >
> > CREATE COLLATION <collation name> FOR
> > <character set specification>
> > FROM <existing collation name>
> > [ <pad characteristic> ]
> >
> > So it seems a collation depends on a character set.
>
> I see. But that really doesn't have anything to do with reality. In
> fact, it completely undermines the transparency of the character set
> encoding that we're probably trying to achieve.

COLLATION being independent of character set is a separate problem
from COLLATION being _defined_ on character set - without a known
character set I can't see how you can define it.
i.e. "COLLACTION for any 8-bit charset" just does not make sense.

-----------------
Hannu