We define a mixin class float-arithmetic-mixin for floating-point arithmetic instructions, and implement register allocation for it. We also ensure that the output of compute-return-value-count-instruction won't be attributed to a register used for return values, and pretend to fix an error in the IR visualizer where compute-dx fails to find an instruction in the sequence of instructions, and make floating-point lexical locations purple to distinguish them easily.
We define a mixin class
float-arithmetic-mixinfor floating-point arithmetic instructions, and implement register allocation for it. We also ensure that the output ofcompute-return-value-count-instructionwon't be attributed to a register used for return values, and pretend to fix an error in the IR visualizer wherecompute-dxfails to find an instruction in the sequence of instructions, and make floating-point lexical locations purple to distinguish them easily.