Thanks for your work on this. As I am not the reporter of the issue and experienced the same problem I have not suspected that it could be my fault.
Sorry for causing trouble.

All the best, Mattin

Right, but this is a error in your code. As I have said already, when Hibernate calls the TRIM SqlFunction it always passes 4 arguments. You are not. Simply fix your code, this is NOT a bug in Hibernate; it is simply your code misusing a Hibernate class.

Partly right, I agree that the provided stack trace is not very helpful here, sorry.

The trim-function is built using the following methods:
public static String buildTrimStringExpression(final Connection connection, final String varName)
throws SQLException
{
final SQLFunction sqlFunction = getSqlFunction(connection, "trim");

return sqlFunction.render(StandardBasicTypes.STRING, Arrays.asList(varName), null);
}

private static SQLFunction getSqlFunction(final Connection connection, final String functionName)
throws SQLException
{
final Dialect dialect = resolveDialect(connection);
final SQLFunction sqlFunction = dialect.getFunctions().get(functionName);
if (sqlFunction == null) {
throw new SQLException("Failed to find '" + functionName + "' function for dialect " + dialect);
}

return sqlFunction;
}

I hope now that you believe me, that the resolveDialect-method correctly returns the HSQLDialect. In fact within this resolveDialect method I'm extending the dialect with the different template:
final Dialect dialect = standardDialectResolver.resolveDialect(metaData);
if (dialect instanceof HSQLDialect) {
// TODO (mah 14.07.2011): patch for issue https://hibernate.onjira.com/browse/HHH-6122
// HSQLDB needs the "both from" keywords in trim function
dialect.getFunctions().put("trim", HSQLDB_TRIM_TEMPLATE);
}

Was I able to explain the problem now?

Best regards, Mattin

In fact, looking at the stack trace that is there... it actually looks like its your code that is preparing the SQL here, not Hibernate...

Thanks for the re-clarification Fred.

Martin, this is not even close to enough information. Your stack trace shows no Hibernate info, so I have no idea where this comes from. As I said before, this is rejected until someone can produce a test case that reproduces this problem. Heck even a simple sample query that you think will reproduce it...