Greg Stark <gsstark@mit.edu> writes:
> Hm. This goes back to the earlier conversation about whether = should ever be
> true for two objects that aren't, well, equal. I thought there was some
> consensus at the time that sorting should impose a superficial ordering on
> items that compare equal but aren't in fact the same.
We forced that recently for text strings (overriding locale-specific
cases in strcoll()), but there certainly was not any intent to decree
that every datatype must do likewise.
regards, tom lane