pow(cpp_dec_float_type(0), cpp_dec_float_type(n)) != 0 for n > 1

[Jan Bouwer <JBouwer@…>]

The overload of 'pow' taking both arguments of number<cpp_dec_float<...>> gives a wrong answer when the base == 0 and the exponent > 1.

eg.

using T = cpp_dec_float_50;
assert( pow(T(0), T(2)) == 0 ); // Failure

Tested against svn #84980 with clang (Apple LLVM version 4.2 (clang-425.0.28) (based on LLVM 3.2svn)) target: x86_64-apple-darwin12.4.0 with: -std=c++11 -stdlib=libc++

[Jan Bouwer <JBouwer@…>]

Demonstration of the error

[Jan Bouwer <JBouwer@…>]

Demonstration of the error - without annoyance, please remove previous

[Jan Bouwer <JBouwer@…>]

Output of the attached code:

pow(0, 0) = 1
pow(0, 1) = 0
pow(0, 2) = 0
pow(1, 0) = 1
pow(1, 1) = 1
pow(1, 2) = 1
pow(2, 0) = 1
pow(2, 1) = 2
pow(2, 2) = 4
pow(0, 0) = 1
pow(0, 1) = 0
pow(0, 2) = 1 - Unexpected!
pow(1, 0) = 1
pow(1, 1) = 1
pow(1, 2) = 1
pow(2, 0) = 1
pow(2, 1) = 2
pow(2, 2) = 4
pow(0, 0) = 1
pow(0, 1) = 0
pow(0, 2) = 1 - Unexpected!
pow(1, 0) = 1
pow(1, 1) = 1
pow(1, 2) = 1
pow(2, 0) = 1
pow(2, 1) = 2
pow(2, 2) = 4
pow(0, 0) = 1
pow(0, 1) = 0
pow(0, 2) = 1 - Unexpected!
pow(1, 0) = 1
pow(1, 1) = 1
pow(1, 2) = 1
pow(2, 0) = 1
pow(2, 1) = 2
pow(2, 2) = 4

[John Maddock]

(In [84997]) Fix bug in eval_pow that causes 0^{N to be none-zero. Fixes #8798.}