Does anyone know if the way you build your query has any effect on its performance?

Example:

String aCatsName = "Tom";
int aCatsAge = 2;
List list = null;

Query query = HibernateHelper.getSession().createQuery("select Cat from Cat where Cat.name = :catname and Cat.age > :catage");
query.setString("catname", aCatsName);
query.setInteger("catage", aCatsAge);
list = query.list();

--OR--

String aCatsName = "Tom";
int aCatsAge = 2;
List list = null;

Query query = HibernateHelper.getSession().createQuery("select Cat from Cat where Cat.name = " + aCatsName + " and Cat.age > " + aCatsAge);
list = query.list();

When you call createQuery does Hibernate initialize anything? When you start "setting" parameters, does Hibernate begin "filtering" the data to work with a smaller subset?

Obviously the first example is "cleaner", I was just curious about how it works under the hood to understand which is more efficient, or if it gets executed the same way.

Thanks for your help.

Brian

As far as hibernate is concerned, there is no difference.

As far as java is concerned, the first option is faster because the string + string approach uses StringBuffer, which is synchronized. Also, java can't put the entire string in its string table.

As far as JDBC is concerned, the first approach is faster, because it creates a PreparedStatement instead of a simple Statement. PreparedStatements are precompiled and stored... somewhere. In driver memory, or in the server, I don't know. Have a look at the JDBC javadocs for more info.

As far as "clean" programming goes, I would recommend putting all queries into .hbm.xml files as named queries. That way. if you ever have to modify a query in use by in-production software, you only have to re-distribute mapping files instead of recompiling entire projects.