RE: Strange result with exponents... -- RPG400-L

| -----Original Message-----
| [mailto:rpg400-l-bounces@xxxxxxxxxxxx]On Behalf Of Jon Paris


| That has to be the answer, although I think you'll find it is a microcode
| issue and not a runtime one as-such.  Coding a test in C or Java should
| confirm that it is not RPG related.
|
| It would be interesting to see if the same difference is noted on
| a box that
| has a hardware floating point unit - if indeed there are such
| animals these
| days.

I was wondering, perhaps along similar lines, how the pSeries worked and
what result would it calculate.  I've *assumed* it had hardware-assisted
numerics, but don't know about floating-point (nor enough about them
a-tall).

| -----Original Message-----
| [mailto:rpg400-l-bounces@xxxxxxxxxxxx]On Behalf Of Hans Boldt

| What changed between releases? Why worry about it at all? The
| difference is
| in the part of the value past the 17 digit. Anything past that
| point cannot
| be trusted with double precision IEEE float format. Ever. That is, as far
| as IEEE float is concerned, if two numbers differ starting at the 18th
| digit, they pretty much represent the same number. The only possible
| difference is that perhaps the algorithm used to format the number in
| readable form has changed. The arithmetic involved almost
| certainly did not
| change. (You could verify that by looking at the float values in hex.)

I think that'd be a Very interesting question, as you said "almost certain"
it's in the display of the formatted number.

| As a rough guideline, I'd simply recommend avoiding float numeric values
| unless you know what you're doing. Perhaps a B (or better) grade
| in senior
| high school physics would be a good pre-req. ;-)

Hans, I got an A in HS physics and read graduate level textbooks to do my
final paper.  I'm NOT stupid, but I'm guessing you know me well enough to
know I do (and say) some of THE most outrageously STUPID things, anyway.
(If I knew why I did some stupid things, I wouldn't...;-)  ANYhoo...

EXACTLY as Joe has recently posted, and from Scott:

| However, I'm quite familiar with the limitations of floating point, I just
| didn't know (until Rob or someone posted it) that the exponent operator in
| ILE RPG outputs floating point.

And some of us may HAVE actually known about this, at one time, and
forgotten.  But Hans, you have the advantage of being primarily concerned
with these kinds of issues, day in and day out.  Application programmers, I
would say, generally think in terms of "I do some arithmetic, and it comes
up with the right answer and then I'm gonna...".  Until something weird
crops up, and it strikes up a curiosity.  (Speaking of which, Scott:  WHY
does DC display THAT many digits in the first place??...;-)

So I'm disagreeing with what Joe said, in part, "Rremember, a lot of us
dinosaurs don't have the benefit of a CS degree,
since when we were coming up there was no such thing."  I can't imagine many
400 programmers don't have the equivalent in work experience.  And I'm
amazed how much coding (though not RPG) that teachers, and journalists and
such can do.  So not having a CS degree, but assuming a more
general-business/social/etc. understanding, shouldn't preclude somebody from
doing business applications, imo.

And, notwithstanding what Buck posted, I'm not sure what is the value of
LIMITING the 400's accuracy to floating-point arithmetic, in the first
place, as it would appear that is also limiting the pSeries in the same way.
You could, apparently, maintain backwards compatibility Very Easily.
Obviously, I don't know the Expense involved in such-a conversion to
all-decimal arithmetic.  (May already Be essentially in place?)

But I do know that a lot of this may depend on Rochester (and others?) and
ToroLabs being on the nearly-same page.  And the folks in Somers being on
the nearly-same page, as well as that being the nearly-same page as
Rochester.  That's a LOTta pages!!  Which is the difficulty in EVERY large
organization, afaik.