John Gunther <mail(at)bucksvsbytes(dot)com> writes:
> In the course of creating a new PG 7.4.3 server, I chose UTF-8 as my
> locale.
UTF-8 is a character set encoding, not a locale. I suspect you may
have chosen en_US or something like that as your locale.
> I now find that sorting is very different with that setting: It
> appears, through trial and error, that all non-alphanumeric characters
> are completely ignored by ORDER BY.
I doubt they are ignored completely, but they probably are ignored in
the first-order comparison. However, this is not PostgreSQL's
bailiwick; we just use whatever strcoll() behavior the C library
provides. The correct place to be looking for this documentation
is in the libc or glibc docs.
regards, tom lane