Today we continue to read the WRL research report on the Swift Java Compiler. We were discussing  runtime check elimination. We said that swift does some simple determinations based on  the users of a value and not based on any hypothesis or theorem. These checks or type tests can be made based on the  properties of the values and the control flow of the method. We saw this for both null check and type tests and the redundancies were removed. The bounds check was a bit more involved. Swift applies condition gathering recursively down to a fixed depth, and applies transitive properties between three variables. Some simple algebraic rules such as sum of two variables are also applied. This approach is effective because Swift's IR provides intermediate access to all the users of a value v, and therefore makes it easy to find program expressions that help prove properties about v. In addition, Swift does a round of global CSE before doing the check removal which lets Swift combine values equivalent to v and collect all their uses together.
We discuss array store checks as another run time check for Swift to optimize. In Java, this check ensures that the object being stored into the array is compatible with the base type of the array. Swift attempts to eliminate array store checks via several methods.  To eliminate any of these checks Swift has to resolve the base type of the array.It will also eliminate an array store check  if the base type of the array is known and can hold the known type of the stored value. If the value is a null or loaded from another element of the array, those checks are eliminated too. Swift also checks the class of the base type of array has no subclasses. If the type is final, then the store check cannot fail and is eliminated.
#coding exercise
GetAlternateOddNumberRangeSumCubeRootTenthPower (Double [] A)
{
if (A == null) return 0;
Return A.AlternateOddNumberRangeSumCubeRootTenthPower();
}