Logically dead code; dubious source code comment / inverted if condition?

[Jonas Maaskola <jonas@…>]

Static code analysis with Coverity pointed me to some dead code in the function ibeta_imp in boost/math/special_functions/beta.hpp. My system's boost library is 1.55.0, but the issue is present identically in boost 1.57. In a block guarded by

if (normalised) {

there is the following line (number 1224):

               fract -= (normalised ? 1 : boost::math::beta(a, b, pol));

Clearly, normalised has to be true due to the if guard thus the second part (boost::math::beta(a, b, pos)) will never be used.

Furthermore, it may be that the if guard's condition is missing a negation for which there are two indications:

• in the vicinity (i.e. in the function ibeta_imp) most other if statements (6 of 7) that involve normalised check if normalised is not set (i.e. "if(!normalised)"), and
• more importantly, the first comment in that block on lines 1211-1213 reads:

  1209:         else if(normalised)
  1210:         {
  1211:            // the formula here for the non-normalised case is tricky to figure
  1212:            // out (for me!!), and requires two pochhammer calculations rather
  1213:            // than one, so leave it for now....
  

I apologize for my limited understanding of the details of the code, which precludes me from proposing a fix, but something seems fishy here!

Finally, the line subsequent to the one with the dead code, i.e. line 1225, contains a commented-out alternative. This indicates to me that this entire block may not have been fully satisfactory to the original author(s).

[John Maddock]

Fixed in ​https://github.com/boostorg/math/commit/20965d162e0b6ae6f3308bceb3bfc5f27d43b031

The logic in the branch is correct - I had originally intenbded this branch to handle both normalized and not cases, but couldn't get the non-normalized case working correctly so it got excluded without tidying up the now-dead code.