使用iBatis开发已经有一年了,这一年来不管愿意不愿意,喜欢不喜欢,都得用,想用Hibernate不行。
随着对iBatis的熟悉,渐渐发现初次接触的人因为不能很好的掌握iBatis使用的方法而对其有一些偏激的看法。
1、iBatis的是用来自由写SQL的,对SQL语句的简单的封装,相当于在JDBC上加了层外壳。
我认为,iBatis的目标是简化JDBC的操作,对SQL的查询结果做了很好的映射封装,这样查询的结果就形成了一个java对象、集合、或者Map等。这样的数据结构相比JDBC结果集的处理是一个质的飞跃。
iBatis虽然支持随意的SQL查询,并且对查询结果集也没有什么特别限制,最懒惰的是定义查询结果集类型为map。但是,这并不是iBatis设计者的本意,我认为,在使用iBatis的时候,尽可能写通用的增删改差sql,然后通过通过这些sql所对应DAO方法的组合,来实现各种业务,而不是写很多特殊的sql、甚至存储过程。当你在更换数据库的时候,移植这些特殊sql和过程就是一件很困难的事情。
2、iBatis映射太麻烦,表字段和实体(Entity,也有人称POJO)对应比较麻烦,不够直观,开发困难。
其实,最好的做法就是使用数据库表字段名作为实体的属性名,这样完全对应起来,问题就解决了。并且,可以写个工具,从数据库生成POJO和SQLMap、DAO、Service等等。这样可以极大提高效率,我就这么做的。
的确如此,对于小数据量是够用了,对于大数据量,这是不可接受的。因此可以通过动态SqL来拼装分页查询Sql。
< dynamic prepend=""
> < isNotNull property ="_start" > < isNotNull property ="_size" > limit #_start#, #_size#
</ isNotNull > </ isNotNull > </ dynamic > 4、iBatis无法做成Hibernate那样的关连查询、级联保存、级联删除。
iBatis可以做到,而且实现很灵活方便。但是比Hibernate操作更容易,也很好理解。这里面对结果集映射类型也很讲究。
< resultMap id ="result_base" class ="Channel" > < result property ="id" column ="id" /> < result property ="stationid" column ="stationid" /> < result property ="remark" column ="remark" /> < result property ="status" column ="status" /> </ resultMap > < resultMap id ="result" class ="Channel" extends ="channel.result_base" > < result property ="cfgList" column ="id" select ="videoregxp.findByChannelId" /> </ resultMap > < resultMap id ="result_base" class ="Videoregxp" > < result property ="id" column ="id" /> < result property ="lmid" column ="lmid" /> < result property ="name" column ="name" /> < result property ="cname" column ="cname" /> < result property ="sortnum" column ="sortnum" /> < result property ="remark" column ="remark" /> </ resultMap > < resultMap id ="result" class ="Videoregxp" extends ="videoregxp.result_base" > < result property ="regex" column ="id" select ="regexpar.findByVideoregxpId" /> </ resultMap > /** * 保存一个视频信息,级联保存相关的剧集和视频分段文件 * * @param videoinfo 视频信息对象 * @return 返回保存后的对象 */ public Videoinfo saveVideoinfoWithCascade(Videoinfo videoinfo) {
videoinfoDAO.saveVideoinfo(videoinfo);
for (Juji juji : videoinfo.getJujiList()) {
juji.setVideoid(videoinfo.getId());
jujiDAO.saveJuji(juji);
for (Secfiles file : juji.getSecfileList()) {
file.setSegmentid(juji.getId());
secfilesDAO.saveSecfiles(file);
}
}
return videoinfo;
}
/** * 删除指定标识的一个栏目 * * @param id 栏目标识 */ public void deleteChannelById(Long id) {
//todo:做可行性校验,以及级联删除等操作,并做业务特殊需求校验 Channel ch = channelDAO.findChannelById(id);
for (Videoregxp v : ch.getCfgList()) {
regexparDAO.deleteRegexparByCfgId(v.getId());
}
videoregxpDAO.deleteVideoregxpByLmId(ch.getId());
channelDAO.deleteChannelById(id);
}
5、iBatis延迟加载好像有问题,本人现在不能确信。
6、iBatis用熟悉了和Hibernate一样容易,通过自己开发工具,可以极大提高开发效率。我会在附件贡献出我的代码生成工具。
7、iBatis的cache缓存没感觉到有多大优势,也没感觉性能的提升。
8、iBatis的动态SQL功能强大,Map参数更能获取最大的灵活性,这是Hibernate和其他持久化框架所无法相比的。
< select id ="findByParams" parameterClass ="map" resultMap ="channel.result" > < include refid ="sql_select" /> < include refid ="sql_where" /> </ select > /** * 根据动态条件查询栏目 * * @param map 参数map * @return 查询结果集 */ public List<Channel> findChannelByParams(Map map) {
return getSqlMapClientTemplate().queryForList(
"channel.findByParams", map);
}
但是map参数也缺乏明确性,使用者无法清楚知道map参数里面究竟可以传什么类型,但是使用者可以通过查看sqlMap进一步获得对参数的详细认识,比如上面的完整SQLMap片段是:
<!--
动态条件分页查询 -->
< sql id ="sql_count" > select count(*)
</ sql > < sql id ="sql_select" > select *
</ sql > < sql id ="sql_where" > from channel
< dynamic prepend ="where" > < isNotNull prepend ="and" property ="stationid" > stationid = #stationid#
</ isNotNull > < isNotEmpty prepend ="and" property ="name" > name like '%$name$%'
</ isNotEmpty > < isNotEmpty prepend ="and" property ="url" > url like '%$url$%'
</ isNotEmpty > < isNotEmpty prepend ="and" property ="regxp" > regxp like '$regxp$%'
</ isNotEmpty > < isNotNull prepend ="and" property ="encoding" > encoding = #encoding#
</ isNotNull > < isNotEmpty prepend ="and" property ="_timeout_ge" > <![CDATA[
timeout >= #_timeout_ge#
]]>
</ isNotEmpty > < isNotEmpty prepend ="and" property ="_timeout_lt" > <![CDATA[
timeout
< #_timeout_lt#
]]
> </ isNotEmpty > < isNotEmpty prepend ="and" property ="_maxrow_ge" > <![CDATA[
maxrow >= #_maxrow_ge#
]]>
</ isNotEmpty > < isNotEmpty prepend ="and" property ="_maxrow_lt" > <![CDATA[
maxrow
< #_maxrow_lt#
]]
> </ isNotEmpty > < isNotEmpty prepend ="and" property ="_start_ge" > <![CDATA[
start >= #_start_ge#
]]>
</ isNotEmpty > < isNotEmpty prepend ="and" property ="_start_lt" > <![CDATA[
start
< #_start_lt#
]]
> </ isNotEmpty > < isNotEmpty prepend ="and" property ="pagereg" > pagereg like '%$pagereg$%'
</ isNotEmpty > < isNotEmpty prepend ="and" property ="kind" > kind = #kind#
</ isNotEmpty > < isNotEmpty prepend ="and" property ="mapname" > mapname = #mapname#
</ isNotEmpty > < isNotEmpty prepend ="and" property ="replaceb" > replaceb = #replaceb#
</ isNotEmpty > < isNotEmpty prepend ="and" property ="replacea" > replacea = #replacea#
</ isNotEmpty > < isNotEmpty prepend ="and" property ="ifredo" > ifredo = #ifredo#
</ isNotEmpty > < isNotEmpty prepend ="and" property ="fenbianlv" > fenbianlv = #fenbianlv#
</ isNotEmpty > < isNotEmpty prepend ="and" property ="_size_ge" > <![CDATA[
size >= #_size_ge#
]]>
</ isNotEmpty > < isNotEmpty prepend ="and" property ="_size_lt" > <![CDATA[
size
< #_size_lt#
]]
> </ isNotEmpty > < isNotEmpty prepend ="and" property ="remark" > remark like '%$remark$%'
</ isNotEmpty > < isNotEmpty prepend ="and" property ="status" > status = #status#
</ isNotEmpty > </ dynamic > < dynamic prepend=""
> < isNotNull property ="_start" > < isNotNull property ="_size" > limit #_start#, #_size#
</ isNotNull > </ isNotNull > </ dynamic > </ sql > < select id ="findByParamsForCount" parameterClass ="map" resultClass ="int" > < include refid ="sql_count" /> < include refid ="sql_where" /> </ select > < select id ="findByParams" parameterClass ="map" resultMap ="channel.result" > < include refid ="sql_select" /> < include refid ="sql_where" /> </ select > 通过上面的映射,我们可以很清楚的看到,SQL的片段得到了重用,分页很优雅,统计和全功能查询也非常强大,相比hibernate,有过之而无不及,当然如果你不会这么用,那就麻烦多了。
9、我认为,iBatis应该尽量使用单表增删改差业务来组合功能更为强大的业务操作。但并不排斥使用多表关联进行查询,多表关联查询,结果集往往需要定义新的JavaBean来进行存储,但也可以简单的返回一个Map来存储结果集。
<!--
动态条件分页查询 -->
< sql id ="sql_count" > select count(a.*)
</ sql > < sql id ="sql_select" > select a.id vid,
a.img imgurl,
a.img_s imgfile,
b.vfilename vfilename,
c.id sid,
c.url url,
c.filename filename,
c.status status
</ sql > < sql id ="sql_where" > From secfiles c, juji b, videoinfo a
where
a.id = b. videoid
and b.id = c.segmentid
and c.status = 0
< dynamic prepend=""
> < isNotNull property ="_start" > < isNotNull property ="_size" > limit #_start#, #_size#
</ isNotNull > </ isNotNull > </ dynamic > </ sql > <!--
返回没有下载的记录总数 -->
< select id ="getUndownFilesForCount" parameterClass ="map" resultClass ="int" > < include refid ="sql_count" /> < include refid ="sql_where" /> </ select > <!--
返回没有下载的记录 -->
< select id ="getUndownFiles" parameterClass ="map" resultClass ="java.util.HashMap" > < include refid ="sql_select" /> < include refid ="sql_where" /> </ select > 在上面,我们的结果集返回为resultClass="java.util.HashMap",这时候,需要根据字
段的名称来取值,值类型为Object,key类型为String,这点要注意了:
public List<Map<String,Object>> findUndownFiles(Map map) {
return getSqlMapClientTemplate().queryForList(
"secfiles.getUndownFiles", map);
}
public void test_findUndownFiles() {
List<Map<String, Object>> co = ser.findUndownFiles(
new HashMap());
StringBuilder s =
new StringBuilder();
for (Map<String, Object> map : co) {
System.out.println(
"---------------------------");
for (Map.Entry<String, Object> entry : map.entrySet()) {
System.out.print(entry.getKey()+
"\t");
System.out.println(entry.getValue());
}
}
}
---------------------------
sid 1
vfilename 200905252009235799
url http://d18.v.iask.com/f/1/f47817a394730dc682e660b943e84cc41006606.flv
status 0
filename 200905252009235799-00.flv
imgfile 200905252009234399.jpg
vid 1
imgurl http://p4.v.iask.com/95/595/1757503_1.jpg
---------------------------
sid 2130
vfilename 2009062615063867492
url http://lz.dhot.v.iask.com/f/1/0ee2ae8b973988f6a93c071c8045ca5217266409.mp4
status 0
filename 2009062615063867492-00.mp4
imgfile 2009062615063825434.jpg
vid 93
imgurl http://cache.mars.sina.com.cn/nd/movievideo//thumb/2/1502_120160.jpg
---------------------------
sid 2131
vfilename 2009062615064184076
url http://lz5.dhot.v.iask.com/f/1/36d3dadacb8d6bda434a58e7418ad3cc19037464.flv
status 0
filename 2009062615064184076-00.flv
imgfile 2009062615064136733.jpg
vid 94
imgurl
10、iBatis在开发效率上并没有比Hibernate差,如果你觉得太难用,那说明你还不够精,两者各有优缺点,应该扬长避短,最终发现:好东西都很简单!希望在这篇文章能为iBatis正名。
本文转自 leizhimin 51CTO博客,原文链接:http://blog.51cto.com/lavasoft/170160,如需转载请自行联系原作者