Jonas Forsman wrote:
> Recreating indicies is maybe not great but can be done quit easily compared
> to reinitializing a db or recoding an entire database.
> Sometimes a bit of a solution is all what it is needed. Would it be
> to change the locale on the fly and only recreate indicies afterwards?
Hmm. I think you could change LC_CTYPE by just hacking the control file.
LC_COLLATE would require the reindexing. Perhaps you could hack that as
well, and then manually run REINDEX on all databases.
Did you confirm what the problem in your case really was? As Greg
pointed out, the example query you posted was in fact flawed..
BTW, 8.3 is more strict on this, and won't allow you to initdb a cluster
with mismatching locale and encoding.
In response to
pgsql-bugs by date
|Next:||From: Euler Taveira de Oliveira||Date: 2007-11-12 12:13:14|
|Subject: Re: BUG #3734: Invalid XML schema output.|
|Previous:||From: Jonas Forsman||Date: 2007-11-12 09:17:16|
|Subject: Re: BUG #3737: lower/upper fails to match extended chars in