Andrus Adamchik pravi:
On Apr 21, 2006, at 12:33 PM, Borut Bolčina wrote:
INFO QueryLogger: DELETE FROM AUTO_PK_SUPPORT WHERE TABLE_NAME IN
('source')
INFO QueryLogger: INSERT INTO AUTO_PK_SUPPORT (TABLE_NAME, NEXT_ID)
VALUES ('source', 200)
An implementation of a smarter auto PK configuration mechanism has
been pending for some time, but I think this particular issue can be
addresses separately. You can try it on a custom PK generator and I
appreciate if you log this issue in Jira. Here is how the solution
might look like:
1. Override JdbcPkGenerator.createAutoPk(DataNode node, List
dbEntities) as follows -
2. Do a select from AUTO_PK_SUPPORT to check which entities are
already present in the DB
3. Subtract those entities from dbEntities list, getting a subset of
entities whose records are missing
4. This line "runUpdate(node, pkDeleteString(dbEntities))" should only
use a subset obtained in (3)
Andrus
The above solution suggests selectively deleting from AUTO_PK_SUPPORT
table which is even more then I want. In my case I have only two tables
plus AUTO_PK_SUPPORT. Once they are created, no deletion in
AUTO_PK_SUPPORT must occur.
If I understand you correctly, the above algorithm would create a statement
DELETE FROM AUTO_PK_SUPPORT WHERE TABLE_NAME IN ('tableA', 'tableB',
'tableC')
but not, say, tableD, as it is already present (created before).
In my case this pkDeleteString would look like (empty table names)
DELETE FROM AUTO_PK_SUPPORT WHERE TABLE_NAME IN ('')
INSERT INTO AUTO_PK_SUPPORT (TABLE_NAME, NEXT_ID) VALUES ('', 200)
I think an error would occur with the above statement.
One "workaround" I can think of is to do a SELECT on AUTO_PK_SUPPORT and
if no error is thrown I must assume the table exists, so I skip
generator.runGenerator(dataSource); altogether.
What do you think?
