Error: "Conversion between UNICODE..."

"Neil Conway" <neilconway@home.com> writes:

I'm getting hundreds of errors in my Postgres (7.0.2) logs like this:
ERROR: Conversion between UNICODE and SQL_ASCII is not supported
Postgres is compiled without multi-byte support and the encoding on the
database is SQL_ASCII. AFAIK, there is no need to store UNICODE (but I
might be wrong -- if that's the problem, tell me).

Since the database is accessed by hundreds of different clients
(interfacing using JDBC, PHP4, Perl, and perhaps others), I'm having
difficulty tracking down where these errors are coming from and why
they're occuring.

Sounds to me like some of your clients *are* compiled with multibyte
support, and are asking for conversion to Unicode. I'm not an expert
on the multibyte stuff though...

regards, tom lane

I'm getting hundreds of errors in my Postgres (7.0.2) logs like this:

ERROR: Conversion between UNICODE and SQL_ASCII is not supported

Postgres is compiled without multi-byte support and the encoding on the
database is SQL_ASCII. AFAIK, there is no need to store UNICODE (but I
might be wrong -- if that's the problem, tell me).

Since the database is accessed by hundreds of different clients
(interfacing using JDBC, PHP4, Perl, and perhaps others), I'm having
difficulty tracking down where these errors are coming from and why
they're occuring.

1. This message is coming from parse_client_encoding() in the backend,
which is never active without multi-byte support. So I'm sure you
are using multi-byte enabled backend. Please check your
installation.

2. The error message indicates that your client asks UNICODE as the
frontend side encoding, but the database is configured with
SQL_ASCII encoding.
--
Tatsuo Ishii