Re: [GENERAL] Invalid unicode in COPY problem

Am Sonntag, den 08.05.2005, 14:30 +0900 schrieb Tatsuo Ishii:
...
> Actually I myself thought as you are before. Later I found that it was
> not so good idea. People already have invalid encoded data in their
> precious database and have very hard time to migrate to newer version
> of PostgreSQL because of encoding validation.
...
> Now I need to upgrade to newer PostgreSQL within 1 day. I know I have
> to fix wrong encoded field but it will take more than 1 day. So I
> would like to import the data first then fix wrong encoded field on
> running database since I can reach members by phone or email even with
> wrong encoded address field...

Well, if you are so in a hurry you better not migrate. Postgres is
proud of validating the input and to have no invalid data. So if
you have invalid data, better fix it.

> I saw this kind of situation in the real world and that's why we
> developed the patches.

Why not developing a helper for contrib to help reencoding the
database instead?

Regards
Tino