Refactor implementation of -Fd flag to make it clear that it's not the F factor which is scaled
In #21483 (closed) we had a discussion about the -Fd
flag and whilst concluding that the behaviour was at least not very wrong, we did agree that the flag is not really scaling the -F
factor at all.
The -F
flag sets the multiplier for how big the heap must get in order to trigger a major collection. By default F = 2
so the heap must grown to 2x the last amount of live bytes before a major collection is triggered.
The -F
flag is used in the calculation about how much memory to retain but really we are not scaling the -F
value because -Fd
doesn't affect how quickly we will decide to do a major collection in the future.
Therefore the implementation could be refactored a little to make it clear that we are not scaling F, even if the ship has sailed for the variable name.