曾几何时......

Once upon a time...

class MusicLibrary {
    private Set<Music> collection ...
    public Set<Music> getPopMusic() { ... }
    public Set<Music> getRockMusic() { ... }
    public Set<Music> getElectronicaMusic() { ... }
}

然后你意识到你想要能够通过其他类型过滤图书馆的馆藏。您可以继续添加新的getter方法。或者您可以使用访客。

Then you realize you'd like to be able to filter the library's collection by other genres. You could keep adding new getter methods. Or you could use Visitors.

interface Visitor<T> {
    visit(Set<T> items);
}

interface MusicVisitor extends Visitor<Music>;

class MusicLibrary {
    private Set<Music> collection ...
    public void accept(MusicVisitor visitor) {
       visitor.visit( this.collection );
    }
}

class RockMusicVisitor implements MusicVisitor {
    private final Set<Music> picks = ...
    public visit(Set<Music> items) { ... }
    public Set<Music> getRockMusic() { return this.picks; }
}
class AmbientMusicVisitor implements MusicVisitor {
    private final Set<Music> picks = ...
    public visit(Set<Music> items) { ... }
    public Set<Music> getAmbientMusic() { return this.picks; }
}

您将数据与算法分开。您将算法卸载到访客实现。您可以通过创建更多访问者来添加功能，而不是不断修改（并膨胀）包含数据的类。

You separate the data from the algorithm. You offload the algorithm to visitor implementations. You add functionality by creating more visitors, instead of constantly modifying (and bloating) the class that holds the data.