From: Pete Bacon Darwin <redmine@...>
Date: 2008-09-23T06:07:17+09:00
Subject: [ruby-core:18796] [Bug #585] BigDecimal#remainder is not correct with small divisors

Bug #585: BigDecimal#remainder is not correct with small divisors
http://redmine.ruby-lang.org/issues/show/585

Author: Pete Bacon Darwin
Status: Open, Priority: Normal

require 'bigdecimal'

# According to the documentation: BigDecimal#remainder is BigDecimal#modulus minus the value divided by, if values have opposite sign

@mixed = BigDecimal("1.23456789")
@pos_int = BigDecimal("2E5555")
@neg_int = BigDecimal("-2E5555")
@pos_frac = BigDecimal("2E-9999")
@neg_frac = BigDecimal("-2E-9999")

# One would expect that 
puts @mixed.remainder(@neg_frac)   == (@mixed % @neg_frac) - @neg_frac    # 0.2E-9998
puts @pos_int.remainder(@neg_frac) == (@pos_int % @neg_frac) - @neg_frac  # 0.2E-9998
puts @neg_frac.remainder(@pos_int) == (@neg_frac % @pos_int) - @pos_int   # -0.2E-9998
puts @neg_int.remainder(@pos_frac) == (@neg_int % @pos_frac) - @pos_frac  # -0.2E-9998

# But in actual fact 
puts @mixed.remainder(@neg_frac)   == BigDecimal("0.0")
puts @pos_int.remainder(@neg_frac) == BigDecimal("0.0")
puts @neg_frac.remainder(@pos_int) == BigDecimal("-0.2E-9998")
puts @neg_int.remainder(@pos_frac) == BigDecimal("-0.0")

# This appears to be due to some premature rounding optimization in divmod


----------------------------------------
http://redmine.ruby-lang.org