You would do this (or close):
<insert id="insert">
INSERT INTO person (lastname, firstname) values (#lastname#, #firstname#)
<selectKey property="personId">SELECT curval('persons_id_seq')</selectKey>
</insert>
After calling the insert method, your bean (or map) would have the
personId property set.
Larry
On 9/30/05, Alan Chandler <[EMAIL PROTECTED]> wrote:
> I am not really sure I understand fully what selectKey is trying to do.
>
> My situation is this.
>
> I have a postgres table "PERSONS" with a primary key of ID which is of type
> SERIAL. In Postgres this means an automatic creation of a SEQUENCE, and a
> default value in the column of nextval('persons_id_seq')
>
> I want to create the transaction to insert a new person into the PERSONS
> table, but I need to know the ID that it has been created with, so that I can
> use that.
>
> There are two options (I think) from a SQL point of view
>
> 1) Within a single transaction
>
> SELECT nextval('persons_id_seq')
>
> - which will output a single value - and which I can pick up in a resultclass
> of int using iBatis, then
>
> INSERT INTO persons (id , ...) values (#id#,...) where id is the value
> returned in the int
>
> 2) Within a single transaction
>
> INSERT INTO persons (data but no the id)
> SELECT curval('persons_id_seq')
>
> and use the value returned from select statement as the ID of the inserted
> row.
>
>
>
> Where does selectKey fit into this - does it perfom the attached select with
> the insert automatically within the same transaction without manually needed
> to define the transaction boundaries? What else does it do?
>
> How to I get back the ID from the API when I call the SqlMap.insert routine.
>
> --
> Alan Chandler
> http://www.chandlerfamily.org.uk
>
