Hi all,
I've had early discussion with Lukas earlier about this, but I'll write the
background first and would like some input from you all.
What I've been trying to get from ORM is to completely separate SQL part
and application code part (dont we all).
I dislike Hibernate's HQL or any other SQL-phobe ORMs and opted for
mybatis. but with mybatis, application code bleeds to SQL part like so:
<mapper namespace="project.electric.asia">
<select id="getAll" resultType="com.test.User">
SELECT * FROM USERS
</select>
<select id="getByCountry" parameterType="string" resultType=
"com.test.User">
SELECT * FROM USERS
LEFT JOIN ADDRESS ON USERS.id = ADDRESS.USER_ID
LEFT JOIN PROJECT ON USERS.id = PROJECT.MEMBER_ID
WHERE id = #{id}
</select>
</mapper>
see that pesky "resultType" attribute there in "select" element? this is
just so not right.
secondly, comes the part where we describe how the flat structured
resultset returned by the query in the above mapper should be mapped to our
graph-like POJO structure
<resultMap id="userResultMap" type="com.test.User">
....
</resultMap>
and bam, the "com.test.User" strikes again. this resultMap xml should be in
plain xml (or json would be nice) and should be used universally by not
only java, but C# or any other OOP language. just like json, we dont need
to put any information about the target object anywhere in the json, so why
couple this xml only to one particular application code?
what I have in mind (in ideal world) things should looks like so
1. the SQL statements (not sure why we use the term "mapper" here. I think
mapper would be more appropriate for the next component). Files like this
are located outside application deployment structure (that's war file if
you're using java)
<mapper namespace="project.electric.asia">
<select id="getAll">
SELECT * FROM USERS
</select>
<select id="getByCountry" parameterType="string">
SELECT * FROM USERS
LEFT JOIN ADDRESS ON USERS.id = ADDRESS.USER_ID
LEFT JOIN PROJECT ON USERS.id = PROJECT.MEMBER_ID
WHERE id = #{id}
</select>
</mapper>
bye bye resultType
2. the resultmap xml (pretty sure we should've used the term "mapper" for
this one here, not the SQL syntax. perhaps because people usually merge
these 2 components into 1 file?). These "mappers" are inside war
deployment, but outside classes folder. I think xml approach is the better
choice to annotating your POJOs like crazy since it reduces noise in your
code and xml is the more loosely coupled way.
<resultMap id="resultMap.namespace.User">
<property objCol="userId" objColType="int" tblCol="user_id"/>
<property objCol="username" objColType="string" tblCol="user_name"/>
<property objCol="address" objColType="complex">
<property tblCol="street_name" objCol="streetName"/>
<property tblCol="street_number" objCol="streetNumber"/>
<property tblCol="country_code" objCol="countryCode"/>
</property>
<property objCol="projects" objColType="complex" isArray="true">
<property tblCol="project_id" objCol="Id" objColType="string"/>
<property tblCol="project_name" objCol="Name" objColType="string"/>
<property tblCol="project_start_date" objCol="startDate" objColType=
"localDate"/>
</property>
</resultMap>
again, bye bye com.test.User
3. and finally the client code will look like this:
List<Users> myTeam = ORMSession.execute("project.electric.asia.getByCountry",
65, "resultMap.namespace.User",User.class);
the second param (65) is the parameter "country code" to be transmitted in
the SQL statement.
for component 1 (SQL statement), Lukas chose the database view option. I
completely forgot about that technology in database and I think it will
work. that solves the first component
now, onto the second component, Lukas recommends model mapper. anybody
knows a good example of this so-called model mapper?
Let me know what you guys think.
Thanks
--
You received this message because you are subscribed to the Google Groups "jOOQ
User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
