BloombergLP::bdlb::Float::isSignalingNan

Description

An invalid floating-point value ("signaling NaN" or "SNaN") is never the result of a valid operation -- it must be produced deliberately (i.e., by calling bsl::numeric_limits<float>::signaling_NaN()). If an SNaN is used in a subsequent operation, the result is undefined and may result in a hardware trap leading to a signal. Some platforms do not support signaling NaNs, especially for single-precision floats, and will convert an SNaN to a QNaN on assignment or copy-initialization (including argument passing). isSignalingNan will always return false on such platforms. Note that, because an SNaN is a NaN, if this method returns true, then isNan(x) will also return true.