"David B" <postgresql@thegatelys.com> writes:
> Two possible bugs.
> #1 Loading unicode data.
I'm thinking that the dump file probably doesn't set client_encoding,
or if it does, sets it to the wrong thing. Is the data in the dump file
actually UTF8, or is it in some other encoding?
> select substring ( lower(title), 1, 1), count(*)
> from album
> where lower(title) > 'a%' and lower(title) <= 'e%'
> group by substring ( lower(title), 1, 1) ;
> Should get counts for 'a', 'b', 'c', 'd', 'e'.
... plus any other characters that the current locale chooses to sort
between 'a' and 'e'. In particular I'd expect to see accented 'a' and/or
accented 'e' sort that way.
regards, tom lane