Eternal loop when assigning infinity or NaN

[PieterB@…]

The following statement gets into an eternal loop in release builds (where NDEBUG is defined):

boost::multiprecision::int128_t inf_i(std::numeric_limits<double>::infinity());

This is caught by a BOOST_ASSERT otherwise. Same is true for NaN.

Question: should the assignment not simply assign 'infinity' instead (or throw an exception)n in such cases?

[John Maddock]

Moved to: ​https://github.com/boostorg/multiprecision/issues/58