Kathey Marsden wrote:
I don't think the tests fork a process or anything so I don't understand how I could be losing the -Xjit:count=0 option. Does anyone have any idea what might be going on?
Most likely the jit is performing a further optimization (ie. a recompile) on the method and producing a correct result. A modern jit does not have a 1-1 mapping between byte code and optimized code.
Dan.
