"Nathan Culwell-Kanarek" <nculwell@wisc.edu> writes:
> Description: 3-octet private use UTF8 chars reported as identical
> We've run into a problem, which is that
> the PostgreSQL backend is interpreting 4 of the private use characters as
> being equivalent.
Your beef is actually with strcoll(); we just believe whatever that
function tells us when comparing strings. Check to see that you've
initdb'd in a utf8-based locale --- if not, that might be the source
of the problem. (IMHO, strcoll ought not claim distinct byte sequences
are equal in any case, but it seems some locale definitions will do
that.) If no luck, take it up with Red Hat's glibc folk.
regards, tom lane
