bug.rb

# Odd behavior in BigDecimal's util for Float#to_d
# on Ruby 1.9.3 and 2.0?
# 
require 'bigdecimal'
require 'bigdecimal/util'

puts "85.26.to_d.to_s('F')              # => #{85.26.to_d.to_s('F')}"
puts "BigDecimal.new('85.26').to_s('F') # => #{BigDecimal.new('85.26').to_s('F')}"

oddities = ('0.00'..'99.99').to_a.select do |v|
  v.to_f.to_d.to_s('F') =~ /\.\d{3}/
end

puts "Found #{oddities.size} oddities:"
puts "--------------------------------"

oddities.each do |v|
  puts "Float: #{v}  #to_d: #{v.to_f.to_d.to_s('F')}"
end

ruby 1.9.3p194 (2012-04-20 revision 35410) [x86_64-darwin12.1.0]

Found 1237 oddities:

Some sample oddities

Float: 65.10  #to_d: 65.09999999999999
Float: 64.46  #to_d: 64.45999999999999
Float: 67.21  #to_d: 67.20999999999999
Float: 75.68  #to_d: 75.68000000000001
Float: 98.26  #to_d: 98.26000000000001

65.10.to_d.to_s('F')  # => "65.09999999999999"

Or more to the point. Should the oddities above be false?

BigDecimal.new('65.09') == 65.09.to_d   # => true
BigDecimal.new('65.10') == 65.10.to_d   # => false
BigDecimal.new('65.11') == 65.11.to_d   # => true

Thats not an oddity, thats the normal problem when you are working with floating point number on a computer..
See http://wikipedia.org/wiki/Floating_point
Just use to_d. With a precision, e.g. 65.10.to_d(10)

Thanks Klaus! I think I may spend some time in ActiveRecord so that value_to_decimal(value) takes a precision so that column reflection can pass down the proper information.

Thanks Ken! Very helpful.