Jira issue originally created by user @beberlei:
* Table ID Generator for those poor languages that are missing sequences.
* WARNING: The Table Id Generator clones a second independent database
* connection to work correctly. This means using the generator requests that
* generate IDs will have two open database connections. This is necessary to
* be safe from transaction failures in the main connection. Make sure to only
* ever use one TableGenerator otherwise you end up with many connections.
* TableID Generator does not work with SQLite.
* The TableGenerator does not take care of creating the SQL Table itself. You
* should look at the `TableGeneratorSchemaVisitor` to do this for you.
* Otherwise the schema for a table looks like:
* CREATE sequences (
* sequence_name VARCHAR(255) NOT NULL,
* sequence_value INT NOT NULL DEFAULT '1',
* sequence*increment*by INT NOT NULL DEFAULT '1',
* PRIMARY KEY (table_name)
* Technically this generator works as follows:
* 1. Use a robust transaction serialization level.
* 2. Open transaction
* 3. Acquire a read lock on the table row (SELECT .. FOR UPDATE)
* 4. Increment current value by one and write back to database
* 5. Commit transaction
* If you are using a sequence*increment*by value that is larger than one the
* ID Generator will keep incrementing values until it hits the incrementation
* gap before issuing another query.
* If no row is present for a given sequence a new one will be created with the
* default values 'value' = 1 and 'increment_by' = 1
* @author Benjamin Eberlei <firstname.lastname@example.org>
Comment created by @beberlei:
Implemented under the DDC-450 label, will get into ORM through this implementation.
Issue was closed with resolution "Fixed"