Bug #5715: +/-1 ** Bignum returns different results than +/-1 ** Fixnum

• Status: Closed
• Assignee: marcandre (Marc-Andre Lafortune)
• Target version: 2.0.0
• ruby -v: ruby 1.9.3p0 (2011-10-30 revision 33570) [x86_64-darwin10.8.0]

[john_firebaugh (John Firebaugh)]

=begin
Consider (({1 ** expt})) and (({(-1) ** expt})).

When ((|expt|)) is a Fixnum, the result is always 1, -1, Rational(1/1), or Rational(-1,1) depending on the signs of the operands and parity of the exponent.

When ((|expt|)) is a Bignum, Float 1.0 is always returned. Either the behavior for Fixnum exponents should be followed exactly (preferred), or at the very least -1.0 should be returned when the base is -1 and the exponent is odd.

$VERBOSE = nil

wordsize = 8 * 1.size
fixnum_max = 2 ** (wordsize - 2) - 1
fixnum_min = -2 ** (wordsize - 2)

[1, -1].each do |a|
[ 1, 2, fixnum_max, fixnum_max + 1, fixnum_max + 2,
-1, -2, fixnum_min, fixnum_min - 1, fixnum_min - 2].each do |b|
puts "%5s ** %20s (%s) == %5s" % [a, b, b.class, a ** b]
end
end

Output:
1 ** 1 (Fixnum) == 1
1 ** 2 (Fixnum) == 1
1 ** 4611686018427387903 (Fixnum) == 1
1 ** 4611686018427387904 (Bignum) == 1
1 ** 4611686018427387905 (Bignum) == 1
1 ** -1 (Fixnum) == 1/1
1 ** -2 (Fixnum) == 1/1
1 ** -4611686018427387904 (Fixnum) == 1/1
1 ** -4611686018427387905 (Bignum) == 1.0 # Bug
1 ** -4611686018427387906 (Bignum) == 1.0 # Bug
-1 ** 1 (Fixnum) == -1
-1 ** 2 (Fixnum) == 1
-1 ** 4611686018427387903 (Fixnum) == -1
-1 ** 4611686018427387904 (Bignum) == 1
-1 ** 4611686018427387905 (Bignum) == -1
-1 ** -1 (Fixnum) == -1/1
-1 ** -2 (Fixnum) == 1/1
-1 ** -4611686018427387904 (Fixnum) == 1/1
-1 ** -4611686018427387905 (Bignum) == 1.0 # Bug
-1 ** -4611686018427387906 (Bignum) == 1.0 # Bug
=end

[john_firebaugh (John Firebaugh)]

The failing cases are the ones that go through the Rational(+/-1) ** Bignum code path, so this is closely related to #5713.

[marcandre (Marc-Andre Lafortune)]

• Category set to core
• Assignee set to marcandre (Marc-Andre Lafortune)

Just saw this update. Yes, as I stated, same issue as 5713. Yes the case for (-1) and bignum exponents can be fixed too. The case for 1 and 0 must be addressed for rational and float exponents too (but -1 will go to float for those)

[shyouhei (Shyouhei Urabe)]

• Status changed from Open to Assigned

[marcandre (Marc-Andre Lafortune)]

• Target version set to 2.0.0

[marcandre (Marc-Andre Lafortune)]

• Status changed from Assigned to Closed
• % Done changed from 0 to 100

This issue was solved with changeset r39063.
John, thank you for reporting this issue.
Your contribution to Ruby is greatly appreciated.
May Ruby be with you.

---

• rational.c (nurat_expt): Deal with special cases for rationals 0, ±1
  [bug #5713] [bug #5715]