hsqldb merge with an IDENTITY column - hsqldb

Question

I have the following hsqldb database:
CREATE TABLE t (id INT IDENTITY, code VARCHAR(10), description VARCHAR(100))
INSERT INTO t (code, description) VALUES ('C1', 'dining table'), ('C2', 'deck chair')
I need to add entries, only if they do not already exist in the table.
I would like to do this in one step (instead of a SELECT step to find out if the entry is in the table, and a second step to INSERT it if that's not the case)
hsqldb has the MERGE operation for this.
however, the problem is that I have a id IDENTITY column that hsqldb should automatically take care of setting the value,
and I don't know how to tell this to the MERGE operation.
MERGE INTO t USING (VALUES(3, 'C3', 'conference table'))
AS vals(x,y,z) ON t.code = vals.y
WHEN NOT MATCHED THEN INSERT VALUES vals.x, vals.y, vals.z
This would work, but I had to give the id (3), and I need that it is hsqldb that automatically sets the id (just as in the INSERT operation above). I would need something like this:
MERGE INTO t (code, description) USING (VALUES('C3', 'conference table'))
AS vals(x,y) ON t.code = vals.x
WHEN NOT MATCHED THEN INSERT VALUES vals.x, vals.y
but this query does not work because the MERGE operation does not allow to specify which columns I am working on "(code, description)".
how can I achieve it?
ps: http://hsqldb.org/doc/2.0/guide/dataaccess-chapt.html#dac_merge_statement

Answer 1

See the INSERT documentation.
You can use the keyword DEFAULT instead of a value. This inserts the default value of a column, or if the column is defined as IDENTITY, the next generated value.

Answer 2

I did not try it but potentially this should work.
This is specification
MERGE INTO <target table> [ [ AS ] <merge correlation name> ]
USING <table reference> ON <search condition>
<merge operation specification>
Now focus on <merge operation specification>. This should be expanded into WHEN MATCHED THEN and/or WHEN NOT MATCHED THEN where you put <merge insert specification>
Which should be regular INSERT statement with. To let HSQL to generate identity you could do 2 things (again theoretical):
Ommit PK column
INSERT [ <left paren> <insert column list> <paren right> ]
[ <override clause> ] VALUES <merge insert value list>
So in your case
MERGE INTO t USING (VALUES(3, 'C3', 'conference table'))
AS vals(x,y,z) ON t.code = vals.y
WHEN NOT MATCHED THEN INSERT (code,description) VALUES vals.y, vals.z
Using identity() method
MERGE INTO t USING (VALUES(3, 'C3', 'conference table'))
AS vals(x,y,z) ON t.code = vals.y
WHEN NOT MATCHED THEN INSERT VALUES identity(), vals.y, vals.z
I will try to test this hypothesis, while I need to something similar.