Re^2: Best Material for Floors: (POSIX--)

[tye]

in reply to Re: Best Material for Floors:
in thread Best Material for Floors:

Bleh. POSIX::floor() on a Perl that uses "long double" (or similar) as NVs induces a great reduction in precision. Much better to just use:

sub floor {  $_[0] < 0  ?  -int(-$_[0])  :  int($_[0]);  }

sub floor { my $i= int( my $f= shift @_ ); return $i - ( $f<0 && $f!=$
+i ); }

Now you know. :)

[lidden]

floor -2.5 Should return -3.

[tye]

Heh, thanks. Fixed. Since I discovered the problem working with really large numbers, even my previous version was tons more accurate than POSIX::floor in that particular situation so it fixed the problem. That's how bad POSIX::floor can be.