When diagnosing issues with mark stack overflows it is very hard to determine a good new initial/maximum mark stack size value. Some output (if not existing) of the maximum high water mark encountered would be really appreciated.