https://bugs.documentfoundation.org/show_bug.cgi?id=163895

b. <[email protected]> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever confirmed|0                           |1

--- Comment #10 from b. <[email protected]> ---
In the end of the day all bin-FP accuracy / rounding issues can be summarized
under "bin-FP issues", but it is better to distinguish the differences between
them such as cancellation, approximated representation, fit-in rounding ... and
to treat them appropriately.  

Similarly, you can group all "libreOffice calc tries prettyfying and can't get
it to work consistently" issues under "unsuccessful prettyfying", which is Mike
Kaganski's level of "There is nothing exact in floating-point calculations in
Calc, ..." ( https://bugs.documentfoundation.org/show_bug.cgi?id=154792 ), but
that does little to change / improve anything.  

This bug is very special or exactly the incontinence to evaluate values when
entering into a table as 16-digit exact, but when writing into a file to save
only 15-digit exact. It only takes a few such inconsistency points and
persistent adherence to them to very effectively block coherent calculations.  

( Partly ) castrating an already weak system (IEEE 754) in order to improve it
is a questionable strategy and leads to even weaker results.

-- 
You are receiving this mail because:
You are the assignee for the bug.

Reply via email to