Have they mentioned yet how fast the java solution is? And how many LOC? Is it even done yet :) ?? Without knowing what data is typical, there's no way to tell which solution is best. Your first solution is the simplest and safest without knowing what the data is (and like you said, it would've been done by now anyway). The FileCache
mentioned earlier or your hash solution might be the best depending on the data. But don't be too disappointed if they don't listen and think perl is "too slow" and don't realize that perl (or some other "scripting language") would've been perfect for this. Also, if they are too perl-inexperienced (and/or java-centric) to realize this, then any perl solution is unmaintainable to them.