https://bugs.documentfoundation.org/show_bug.cgi?id=162330

--- Comment #4 from Mike Kaganski <[email protected]> ---
FTR: the function doing actual approximate comparisons in Calc is

https://opengrok.libreoffice.org/xref/core/sal/rtl/math.cxx?r=0ccf4478&mo=24361&fi=629#629

and it compares the absolute value of calculated difference (~7.1*10^-14,
~1.4*10^-14, ~8.5*10^-14) with the result of multiplying absolute value of
arguments, i.e. ~22.67, by 2^-48 (which is ~8.05*10^-14 in this case). The
explanation is, that the comparison only takes 48 significant bits of the
compared numbers into account, out of 53 - the last 5 bits are considered to be
unreliable. This is the function that is already built into Calc, to protect
the user against too aggressive exposition to the imprecision of computer
floating-point arithmetics. As seen, the last difference's absolute value is
greater than the threshold.

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

Reply via email to