Tom Lane wrote:
spellberg_robert <emailrob@xxxxxxxxxxxx> writes:
i was wondering: what was it ?
further, is this, per_chance, described, in excruciating detail,
in a location where i did not tread ?
Did you read the 8.3.3 release notes?
http://www.postgresql.org/docs/8.3/static/release-8-3-3.html
regards, tom lane
hi, tom ---
no. i hadn't.
i just did.
at the risk of ridicule, i don't read release notes online.
i read them when i unpack a tarball.
the first place i tried was the -bugs list for jun_11, the ftp date for 8.3.3.
then, i worked back in time for a couple of weeks.
not finding anything there, i tried -general.
at that point,
i decided the better course of valor was to ask someone to point the way.
otherwise, i would have tried another list.
i had a feeling it would be someplace "obvious".
per mr. marlowe's suggestion, i was perusing -hackers.
i had just found your original post of jun_06 when your e_mail arrived.
i thank you and him.
while i was reading -hackers, i came across justin's post of jun_07.
i've dealt with this aggravation.
i don't consider myself to be qualified to post to -hackers, yet,
so, i'll toss my two u_s_a_denominated copper pieces into this ring.
physically, regardless of currency, money is an integer thing.
use a floating_point representation --only-- for intermediate results
that involve log [ or trig, etc. ] calculations
[ e. g., in a present_value analysis ].
once you have the final answer, convert it into an integer.
in usa terms, the unit may be dollars, centi_dollars or micro_dollars,
as three examples, according to your application.
what is important is that it be stored as an integer.
then, you may safely add thousands of them in a loop.
floating_point round_off error has been a pet peeve of mine for decades.
hope this helps.
rob
