Message-Id: <200009140339.UAA01240@dmz0.etranslate.com>
To: pgsql-interfaces@postgresql.org
Subject: bug in JDBC UNICODE support
From: "Eric D. Friedman" <eric@etranslate.com>
Reply-To: "Eric D. Friedman" <eric@etranslate.com>
Date: Wed, 13 Sep 2000 20:38:45 -0700


I built a postgresql.jar file from the HEAD of CVS on 9/13/2000.
According to the CVS logs, this includes code to support correct
handling of UNICODE (UTF-8) characters in the JDBC driver, provided
that the database is, in fact, a unicode database.

The driver simply does not work as advertised.  I ran the
example.Unicode test and it confirmed the problem I encountered in my
own code by reporting that each of the strings inserted was different
from those returned by a subsequent select.

Browsing in the pgsql-interfaces archives, I found that a
user-contributed patch that DOES work and which has a bit more
intelligence about setting reasonable defaults for the encoding (it
asks the backend which encoding to use).

That patch is available at

http://www.se.postgresql.org/mhonarc/pgsql-interfaces/2000-09/msg00038.html

I believe that this patch - which is a superior implementation, I might
add - should replace the current, broken encoding handler.

Eric

P.S.  I tried to submit this as a bug using the web interface for bug
reports, but that reported the following error:

Warning: Unable to connect to PostgreSQL server: No pg_hba.conf entry for host localhost, user vev, database postgresql in
/home/projects/pgsql/ftp/www/html/bugs/bugs.php on line 150
Unable to access database

It is unclear to me how to report a bug on the bug system. ;-)