Java has an option, -XX:MaxRAMFraction=XXX, that lets developers specify the maximum amount of available RAM a Java application should occupy. This sounds like a great idea, until you realize how it’s implemented. The number that you specify for “XXX” is an integer that represents the fraction of total RAM that should be used. The intrepretation is 1/XXX. So -XX:MaxRamFraction=1 tells the application to use as much RAM as it wants. XX:MaxRAMFraction=2 tells it to use half of the memory. What if you want to use 75% of the memory? Or, perhaps 2/3 the memory? Tough luck. You get all, one-half, one-third, one-fourth, one-fifth, etc. You want to use 40%? Sorry, buddy. You can get 50% or 33%. Take your pick.

When I first heard this, I thought the guy who was explaining it was misinformed or pulling my leg.

I’d sure like to understand the thinking that went into this design decision, because from where I’m standing it looks like lunacy. I can’t imagine how such an idiotic design could make it through review and implementation without somebody with a modicum of intelligence and authority raising a stink. You mean nobody realized how stupid this is?

Java 10 introduced the -XX:MaxRAMPercentage and associated flags. See https://bugs.openjdk.java.net/browse/JDK-8186315 for more information.

Sanity prevails, but I still want to understand how the original design got approved. That something so obviously wrong managed to make it through the approval process and was released on an unsuspecting world doesn’t give me much confidence in the competence of Java platform developers.