Running Jasper Reports against an in-memory h2 datasource? - postgresql

Question

I'm trying to run jasper reports against a live and reporting database, but any reports run against the live database throw exceptions about not finding the right tables (although the default PUBLIC schema is found). It looks like the main DataSource connection isn't honoring the H2 connection settings which specify IGNORECASE=true, as the generated columns and tables are capitalized, by my queries are not.
DataSource.groovy dataSource:
dataSource {
hibernate {
cache.use_second_level_cache = false
cache.use_query_cache = false
}
dbCreate = "create-drop" // one of 'create', 'create-drop','update'
pooled = true
driverClassName = "org.h2.Driver"
username = "sa"
password = ""
url = "jdbc:h2:mem:testDb;MODE=PostgreSQL;IGNORECASE=TRUE;DATABASE_TO_UPPER=false"
jndiName = null
dialect = null
}
Datasources.groovy dataSource:
datasource(name: 'reporting') {
environments(['development', 'test'])
domainClasses([SomeClass])
readOnly(false)
driverClassName('org.h2.Driver')
url('jdbc:h2:mem:testReportingDb;MODE=PostgreSQL;IGNORECASE=TRUE;DATABASE_TO_UPPER=false')
username('sa')
password('')
dbCreate('create-drop')
logSql(false)
dialect(null)
pooled(true)
hibernate {
cache {
use_second_level_cache(false)
use_query_cache(false)
}
}
}
What fails:
JasperPrint print = JasperFillManager.fillReport(compiledReport, params,dataSource.getConnection())
While debugging, the only difference I've found is that the live dataSource, when injected or looked up with DatasourcesUtils.getDataSource(null), is a TransactionAwareDatasourceProxy, and DatasourcesUtils.getDataSource('reporting') is a BasicDataSource
What do I need to do for Jasper to operate on the active in-memory H2 database?
This failure is not reproducible against a real postgres database.

Answer 1

Probably you are opening a different database. Using the database URL jdbc:h2:mem:testDb will open an in-memory database within the same process and class loader.
Did you try already using a regular persistent database, using the database URL jdbc:h2:~/testDb?
To use open an in-memory database that is running in a different process or class loader, you need to use the server mode. That means, you need to start a server where the database is running, and connect to it using jdbc:h2:tcp://localhost/mem:testDb.
See also the database URL overview.

Answer 2

H2 doesn't currently support case-insensitive identifiers (table names, column names). I know other databases support it, but currently H2 uses regular java.util.HashMap<String, ..> for metadata, and that's case sensitive (whether or not IGNORECASE is used).
In this case, the identifier names are case-sensitive. I tried with the database URL jdbc:h2:mem:testReportingDb;MODE=PostgreSQL;IGNORECASE=TRUE;DATABASE_TO_UPPER=false using the H2 Console:
DROP TABLE IF EXISTS UPPER;
DROP TABLE IF EXISTS lower;
CREATE TABLE UPPER(NAME VARCHAR(255));
CREATE TABLE lower(name VARCHAR(255));
-- ok:
SELECT * FROM UPPER;
SELECT * FROM lower;
-- fail (table not found):
SELECT * FROM upper;
SELECT * FROM LOWER;
So, the question is: when creating the tables, were they created with uppercase identifiers or a different database URL? Is it possible to change that? If not: is it possible to use a different database URL?

Answer 3

Just don't run reports against in-memory datasources, and this won't be an issue.