Version: (using KDE KDE 3.5.0) Installed from: Compiled From Sources In the FAQ section of kcalc's manual, there's a comment about the calculator application of Windows. It says that this application uses the same IEEE arithmetic as kcalc and thus has the same rounding errors. This is not true anymore; calc.exe uses an arbitrary precision library. Read more about it in this blog post: http://blogs.msdn.com/oldnewthing/archive/2004/05/25/141253.aspx
kcalc uses an arbitrary precision library since KDE 3.5.0 too. :-)
SVN commit 509194 by annma: remove obsolete doc BUG=120812 M +1 -10 index.docbook --- branches/KDE/3.5/kdeutils/doc/kcalc/index.docbook #509193:509194 @@ -650,17 +650,8 @@ <answer><para>The main factor determining the precision of &kcalc; is whether your libc and libmath supports the C data type <type>long double</type>. If this is the case, &kcalc; will detect this at compile time and use it as its fundamental data type to represent numbers. -Check &kcalc;'s <guilabel>About</guilabel> dialog (in the -<guibutton>Configure</guibutton> dialog box) in order to find out what the -fundamental data type for your copy of &kcalc; is.</para> +</para> -<para>Unless you have a libc and libmath of exceptionally high quality, you will -be able to observe some <quote>interesting</quote> results when trying -to execute computations such as: <userinput>123.22 - 123.21</userinput>, -<userinput>2.01 - 2</userinput>, <userinput>123.88 - 123.87</userinput> and -similar. However if you think this is bad I ask you to do the same computation -on the calculator provided with &Windows;.</para> - <para>Adjust the <guilabel>Precision</guilabel> in &kcalc;'s <guibutton>Configure</guibutton> dialog so that the above computations work correctly. I recommend a precision of 14 if the fundamental data type