Thanks for explaining this Geoff. My math is somewhat basic, so it's always good to learn new things like this. On 29 Jun 2011 08:00, "Geoff Kuenning" <ge...@cs.hmc.edu> wrote: > > > On Jun 29, 1:37 am, Tom Evans <tevans...@googlemail.com> wrote: >> Damn, this was the bit I meant to comment upon - hard and fast rules >> are dangerous. Integral types are distinctly different to floating >> point types, and you should be aware of which one you are using and >> why. You definitely should not be using floats when you require >> integral mathematics, or reliable accounting since 1.0 is only an >> approximation to 1. >> >> Cheers >> >> Tom >> >> >>> 1.0 * 10000000000000000000000000000 >> >> 9.9999999999999996e+27 > > Actually, 1.0 is NOT an approximation to 1. It is precisely 1. All > integral values up to 2^53-1 can be represented exactly. So can many > rational fractions (roughly, those in which the denominator is a power > of two and the numerator and denominator don't differ by a really big > ratio). > > The reason the above example fails isn't the 1.0, it's because the > second number (10^28) is bigger than 2^53, so IT can't be represented > exactly. > > -- > You received this message because you are subscribed to the Google Groups "Django users" group. > To post to this group, send email to django-users@googlegroups.com. > To unsubscribe from this group, send email to django-users+unsubscr...@googlegroups.com. > For more options, visit this group at http://groups.google.com/group/django-users?hl=en. >
-- You received this message because you are subscribed to the Google Groups "Django users" group. To post to this group, send email to django-users@googlegroups.com. To unsubscribe from this group, send email to django-users+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/django-users?hl=en.
