java1.5引入了annotation，初看起来非常cool，但是仔细想想它可以被用在什么地方呢？什么样的需求才是用annotation的最佳方式呢？

偶先举几个例子吧：
1. O/R Mapping
比如EJB3的持久化映射, 具体就不用说了， 规范里面非常清楚了, 看代码：

@NamedQueries({
    @NamedQuery(
        name="authorPosts",
        queryString="select p from Post p where p.author = :author"
    );,
    @NamedQuery(
        name="allPosts",
        queryString="select p from Post p"
    );
});

@Entity
@Table(name="FORUM_POST", schema="javaeye");
public class Post {
    private Long _id;
    private String _title;
    private String _content;
    private User _author;
    
    @Id(generate=SEQUENCE, generator ="POST_SEQ");    
    public Long getId(); {
        return _id;
    }
    
    @Column(name="TITLE", nullable=false, length=255);
    public String getTitle(); {
        return _title;   
    }
    
    @Column(name="CONTENT", columnDefinition="CLOB NOT NULL");
    public String getContent(); {
        return _content;
    }
    
    @ManyToOne(optional=false);
    @JoinColumn(name="AUTHOR_ID", nullable=false, updatable=false);
    public User getAuthor(); {
        return _author;
    }
}

2. 校验
比如WebWork的visitor方式校验，能够将action的校验，转发到Entity上, 以前的写法是写一个Post-validation.xml:

<!DOCTYPE validators PUBLIC "-//OpenSymphony Group//XWork Validator 1.0//EN" "http://www.opensymphony.com/xwork/xwork-validator-1.0.dtd">
<validators>
	<field name="title">
        <field-validator type="requiredstring">
            <message>Please input title.</message>
        </field-validator>
    </field>
    <field name="content">
        <field-validator type="requiredstring">
            <message>Please input content.</message>
        </field-validator>
    </field>
</validators>

改用annotation成:

    @Validator(type="requiredstring", message="Please input title.");
    public String getTitle(); {
        return _title;   
    }
    
    @Validator(type="requiredstring", message="Please input content.");
    public String getContent(); {
        return _content;
    }

3. 索引
比如用lucene做全文搜索，以前是写一个Post-lucene.xml, 当一个Post被持久化的时候，能够自动根据这个配置文件对相应的属性做不同的索引处理:

<lucene>
    <field type="Keyword" fieldName="title"/>
    <field type="Text" fieldName="content"/>
</lucene>

改用annotation成:

    @LuceneIndex(type="KEYWORD");
    public String getTitle(); {
        return _title;   
    }
    
    @LuceneIndex(type="TEXT");
    public String getContent(); {
        return _content;
    }

如果将3种annotation都混合在一起，整个Post Entity将会变成：

@NamedQueries({
    @NamedQuery(
        name="authorPosts",
        queryString="select p from Post p where p.author = :author"
    );,
    @NamedQuery(
        name="allPosts",
        queryString="select p from Post p"
    );
});

@Entity
@Table(name="FORUM_POST", schema="javaeye");
public class Post {
    private Long _id;
    private String _title;
    private String _content;
    private User _author;
    
    @Id(generate=SEQUENCE, generator ="POST_SEQ");    
    public Long getId(); {
        return _id;
    }
    
    @Column(name="TITLE", nullable=false, length=255);
    @Validator(type="requiredstring", message="Please input title.");
    @LuceneIndex(type="KEYWORD");
    public String getTitle(); {
        return _title;   
    }
    
    @Column(name="CONTENT", columnDefinition="CLOB NOT NULL");
    @Validator(type="requiredstring", message="Please input content.");
    @LuceneIndex(type="TEXT");
    public String getContent(); {
        return _content;
    }
    
    @ManyToOne(optional=false);
    @JoinColumn(name="AUTHOR_ID", nullable=false, updatable=false);
    public User getAuthor(); {
        return _author;
    }
}

本来干干净净的一个Post POJO，变成了充满annotation的怪家伙，这可以称之为annotation的滥用么？难道又一个AOP出现了？难道以后偶们打开一个Java文件必须用Eclipse之类的IDE，好让它自动将annotation fold，免得花了眼睛么？

你对annotation的意见如何？