Thread: BUG #1811: unique index problem

The following bug has been logged online:

Bug reference:      1811
Logged by:          George Jansen
Email address:      gjansen@aflcio.org
PostgreSQL version: 8.0.3
Operating system:   RedHat  ES 4
Description:        unique index problem
Details:

Importing a database in preparation for a Lyris cutover, I received an error
creating a unique index. In psql I checked the table:

lyris=# CREATE UNIQUE INDEX ix_uniquewords_word ON uniquewords_ USING btree
(word_);
ERROR:  could not create unique index
DETAIL:  Table contains duplicated values.
lyris=# SELECT word_, Count(*)
lyris-# FROM uniquewords_
lyris-# GROUP BY word_
lyris-# HAVING Count(*) > 1;
 word_ | count
-------+-------
(0 rows)

lyris=#


Then I listed the words out in order to a text file
and ran


-bash-3.00$ uniq -c uwords | grep -v "^ *1"
      2  avendano
      2  espanol
-bash-3.00$

Both old and new databases use SQL_ASCII. Is this a bug?

"George Jansen" <gjansen@aflcio.org> writes:
> [ inconsistent comparison behavior ]
> Both old and new databases use SQL_ASCII. Is this a bug?

I think it's probably a locale problem; did you initdb the database in C
locale?  (Check SHOW LC_COLLATE and SHOW LC_CTYPE to be sure.)  Usually
we see this sort of thing when the locale is expecting characters in a
different encoding than what's actually being stored.

            regards, tom lane