IMHO this is a bad way or trying to reduce code.
Let me point out a few reasons :)
1. Why use a map if you are only ever going to pass 1 string in. That is
twice the amount of memory needed to object creation.
2. You have moved database specific information into your java code. I
think it is a good practice to have all sql in the maps. I think it is much
more readable and easier to debug. You java code will be littered with
map.get("SOME_COLUMN") calls. Not clean in my opinion.
3. As long as you keep the static table name string in your DAO layer you
should not run into SQL injection but I get scared every time I see the $$.
4. Some one might correct me but I believe that using actual result classes
with result maps is more performant then always using maps.
Sorry to be so negative, I just woke up :)
Nathan
On 2/28/07, Stefano Mancini <[EMAIL PROTECTED]> wrote:
Hi,
I've the following problem:
I want to read the rows of a generic table into a List of HashMaps using a
statement like this one
<select id="selectAll" remapResults="true" resultClass="java.util.HashMap"
parameterClass="java.util.Map">
select * from $table$
</select>
If the table contains a column of type BLOB in the resulting map is
inserted an object of type oracle.sql.BLOB (I'm using oracle 10g), I would
like to have a byte[] in this case. I've inserted the following mapping
<typeHandler jdbcType="BLOB" javaType = "[B" callback="
com.ibatis.sqlmap.engine.type.BlobTypeHandlerCallback" />
in the SqlMap with no results (I've used "[B] as javaType because byte[]
isn't accepted).
Obviously if I use a resultMap with all the properties specified all works
fine, but it's no more general.
Any ideas ?
