为什么Java编译器不喜欢原始int作为HashMap中值的类型？

Question

The compiler complains about this code:

    HashMap<String,int> userName2ind = new HashMap<String,int>();
    for (int i=0; i<=players.length; i++) {
        userName2ind.put(orderedUserNames[i],i+1);
    }

It writes "unexpected type" and point on int . If I replace int by String and i+1 by i+"1" , the compilation goes OK. What is wrong with in here?

Answer 1

It's fine with Integer , but not okay with int - Java generics only work with reference types , basically :(

Try this - although be aware it will box everything:

HashMap<String,Integer> userName2ind = new HashMap<String,Integer>();
for (int i=0; i<=players.length; i++) {
    userName2ind.put(orderedUserNames[i],i+1);
}

Answer 2

If you have small collections, then using reference types is probably fine, but there are alternatives and good one is trove4j . Trove does a pretty good job of recreating the collections API using pure primitives. The payoff is much lower memory usage and in many cases, better performance when inserting/looking up. Your example would look like this:

TObjectIntHashMap<String> userName2ind = new TObjectIntHashMap<String>();
for (int i=0; i<=players.length; i++) {
    userName2ind.put(orderedUserNames[i],i+1);
}

The only downside, in my experience, is the absence of concurrent implementations of these, so you have to figure out another way to manage thread safety.