Thomas Lockhart <lockhart@alumni.caltech.edu> writes:
> How about adding some regression test queries to catch this kind of
> thing? Looks like we don't have *anything* in this area at all except
> for tests in the multi-byte string handling, from Tatsuo.
Yeah, I was thinking the same thing. I'll bet the MB tests don't catch
this bug either, because it's substantially less likely to get noticed
if USE_LOCALE is on...
regards, tom lane