From: "Jesús Gabriel y Galán" <jgabrielygalan@...>
Date: 2010-04-07T20:07:25+09:00
Subject: Re: Precision and ==

On Wed, Apr 7, 2010 at 1:02 PM, LAMBEAU Bernard <blambeau@gmail.com> wrote:
> We all know that finite representation comes with some difficulties...
> but...
>
> x = 45.0*(Math::PI / 180.0)
> y = 0.785398163397448
> puts "#{x.class} : #{x}"
> puts "#{y.class} : #{y}"
> puts 45.0*(Math::PI / 180.0) == 0.785398163397448
>
> prints
>
> Float : 0.785398163397448
> Float : 0.785398163397448
> false
>
> Does to_s show something less precise than the internal representation of
> Floats?

Yes, it rounds to 15 decimal positions:

irb(main):001:0> x = 45.0*(Math::PI / 180.0)
=> 0.785398163397448
irb(main):002:0> y = 0.785398163397448
=> 0.785398163397448
irb(main):003:0> puts "#{x.class} : #{"%.20f" % x}"
Float : 0.78539816339744827900
=> nil
irb(main):004:0> puts "#{y.class} : #{"%.20f" % y}"
Float : 0.78539816339744794593

Jesus.