[Libreoffice-bugs] [Bug 132083] FORMATTING - Wrong date substraction in cell's operation

[bugzilla-daemon at bugs.documentfoundation.org]

https://bugs.documentfoundation.org/show_bug.cgi?id=132083

--- Comment #11 from Mike Kaganski <mikekaganski at hotmail.com> ---
(In reply to jarko from comment #10)
> The whole discussion is about what kind of method is preferred when
> presenting data: impossible exact binary representation of given number or
> exact number known to user. If program/procedure/script assumes that "user
> has always right", than it proceed with any operation needed to finally get
> that exact number. If not - simply do the best with arithmetic, but with
> rounding error.

Sigh. You seem to not follow the "discussion".

First, you confuse "presenting data" and storing the data. Then you mix the
specific issue raised here in this tdf#132083 (which is about changed *display*
- presentation - of the very same data that was already in previous versions)
with an irrelevant "discussion" about "any operation needed to finally get that
exact number" (this is not about representation, but about calculations), which
is explained in the FAQ. In the end, you don't realize that there's *no* way to
do what you think is doable ("any operation needed to finally get that exact
number"). Very often, simple (for a basic school pupil) mathematics would need
infinite time and memory to be calculated exactly in existing computers. So
this is not some meaningful "discussion" about what "kind of method" to use,
but rather about "what this world is" vs "what I want it to be", which is
absolutely useless in this issue - it's simply offtopic. You *can not* make
calculations with data with infinite precision. You can't even have a true Pi
in a modern computer, no matter how long will you calculate that to "assume
that user has always right".

*Sometimes* some *small subset* - not all - of operations could be calculated
exactly if done differently (it means, not using the computer's arithmetic
processor, but with in software algorithms requiring much extra memory, and
much extra time). But even that is *not* "assuming that user has always right",
because the primary goal of spreadsheet software is *massive calculations* on
big amounts of data represented in tabular form. The user for whom the software
is created needs it to process those big amounts in timely fashion, and it's
*that* user who the software assumes is right in that expectation. If Calc
would try to give exact results even in theoretically possible cases, its users
(who use it for real job, not as a calculator for two numbers) would say "It
must give us answers, not f**k up our precious time for own enjoyment
calculating with full precision, which taking years to complete".

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/libreoffice-bugs/attachments/20200414/af941140/attachment.htm>