Re: BUG #3737: lower/upper fails to match extended chars in LATIN1

(please keep the list cc'd)

Jonas Forsman wrote:
> if I cannot change charset within a postgres installation
> and then benefit from all the features in postgres isn't that to be seen
> as a bug
> or at least it should be a workaround easier than dumping everything
> which I think is a too large risk to take if not really neccessary.

Well, it's not a bug in the sense that it's documented and working as
designed.

But I agree that it sucks. Unfortunately changing the locale on-the-fly
isn't as easy as it might sound. The locale affects the sort-ordering,
for example, which means that you'd have to at least recreate all
indexes with text data in them anyway.

Changing the character encoding would be even worse. You'd have to
re-encode all data, which means rewriting all data in the database.

The best solution would be to support per-database and even per-table
and per-column locales and character encodings, so that you could do the
conversions piece by piece. Patches are welcome :-).

--
Heikki Linnakangas
EnterpriseDB http://www.enterprisedb.com