- https://en.wikipedia.org/wiki/Rounding#Floating-point_rounding
- https://en.wikipedia.org/wiki/Rounding#Round_half_away_from_zero
- https://en.wikipedia.org/wiki/printf_format_string
- http://www.gnu.org/software/libc/manual/html_node/Rounding.html
- http://stackoverflow.com/questions/178539/how-do-you-round-a-floating-point-number-in-perl
To be more precise: I actually mean “considering a given precision” of real numbers.
In Germany (in a financial context at least) “round half away from zero” is being used. …
Look at this – can you explain it?:
$ perl -e '$h = 3272.505; printf "%.2f --- %.10fn",$h,$h;'
3272.51 --- 3272.5050000000
$ perl -e '$h = 622.775; printf "%.2f --- %.10fn",$h,$h;'
622.77 --- 622.7750000000
I think we don’t see, what the real numbers really are. 3272.505 is supposedly kept as something like 3272.50499999 (<) as supposed to 622.775, which is supposedly kept closer (>=) to what we see. Apparently the results are acceptable outside the financial world, but within the financial world they are totally unacceptable. I have no idea, why the Perl community has been allowing this for such a long time.
Update 2015-11-21:
I fiddled on a subroutine, that essentially takes parameters like this:
sprintf("%7.2f",3272.505);
For the time being I call it like this:
&round_real_number__half_away_from_zero(
'v' => 3272.505,
'total_length' => 7,
'post_decimal_points_digits' => 2);
In a way I am quite disappointed, that sprintf does not do, what I expect it to do.