Yes, it really depends on Locale. If I created a database with Locale = C, the problem won't happen (at least so far). BTW, I forgot to mention I tested with some Chinese and Japanese characters.
However, it raises another issue why a wrong Locale would damage index? It shall only affect records with wrong conversion, not something like '1000'. The real environment in my program is J2EE+JDBC. I simply used pgAdmin to demonstrate the problem (not caused by JDBC or so). Though Locale C solved the problem, LIKE 'prefix%' still uses Seq Scan. It is a job to change hundreds (if not thousands SQL) to use BETWEEN/AND instead. Any suggestion? -----Original Message----- From: Dave Page [mailto:[EMAIL PROTECTED] Sent: Tuesday, February 22, 2005 12:19 AM To: Richard Huxton Cc: Tom Yeh; pgsql-bugs@postgresql.org Subject: RE: [BUGS] BUG #1487: Index problem > -----Original Message----- > From: Richard Huxton [mailto:[EMAIL PROTECTED] > Sent: 21 February 2005 16:13 > To: Dave Page > Cc: Tom Yeh; pgsql-bugs@postgresql.org > Subject: Re: [BUGS] BUG #1487: Index problem > > > pgAdmin uses libpq, just as psql does, so I cannot imagine why this > > would be the case. pgAdmin also does not do anything to try > to affect > > query plans in any way. > > Could it set the encoding/locale differently to psql (on the > same machine)? Good point, yes - it will try to set the encoding to unicode if it can. /D __________________________________ Do you Yahoo!? The all-new My Yahoo! - Get yours free! http://my.yahoo.com ---------------------------(end of broadcast)--------------------------- TIP 3: if posting/reading through Usenet, please send an appropriate subscribe-nomail command to [EMAIL PROTECTED] so that your message can get through to the mailing list cleanly
