unicode sorting problems

Hello!

I'm having problems with sort order of special characters in a db
cluster which contains LATIN2 and UNICODE databases.

If I initialise the cluster without a locale or with sl_SI.utf, then
only the sort order for unicode databases is correct and special chars
end up in wrong places in latin2 databases - and the other way around
when using sl_SI as cluster locale.

Is there a way to get correct sorting order in both cases, other than
running two separate servers with two separate clusters?

Is creating all databases with unicode encoding and setting
client_encoding to latin2 (where needed) a valid solution to this
problem? Will there be a noticable delay when reading data?

I'm using Postgres 7.3.1

Thank you.

Best regards,
Aleksander