Re: Re: Why is the comparison between timestamp and date so much slower then between two dates

Thomas Kellerer <spam_eater@gmx.net> writes:
> So the optimizer _should_ be smart enough to do the conversion only
> once at the beginning of the statement and then use that converted
> value during the execution of the statement without the need
> to re-evaluate it for each row.

It's not; especially not in your originally posted case where the
up-conversion happens on the variable not the pseudo-constant.

            regards, tom lane