DAO, also called mapper (I think), is a Core Java EE pattern.
I love this pattern since (IMO) creating a layer of abstraction between code and SQL increases a lot the quality of the code.
I usually follow the Java EE pattern, but when you need some really specific data from the DB, I feel a little bit lost. I am not sure who should be responsible for retrieving that data.
My approach is usually to create a UserModel class, in which I store the most used parameters, and in the DAO I create a method getUser(param: String): UserModel. But, what if there is a column in the DB which of course is needed, but not so often. What is the best approach?
Do you think DAOs should provide methods like (e.g.) getUserID(): String or isUserVerfied(): Boolean? or do you think the responsibility of the DAO is just providing the basics (update/delete/insert/select) methods?
Basically, the question is, when you need some data which is not used a lot in your code, what is the best approach, create a select() method in your DAO which loads everything (of users’ table (e.g.)) in your model. Or having some specific method like isUserVerified(): Boolean despite this is not part of the pattern? Or there is another approach which works in both situations (often use fields and not often used fields)?
Please note that the answer should be the most language-independent as possible.
The purpose of a DAO is to hydrate complete objects, and in most cases it shouldn’t be a problem to always load all of an object’s fields. Not loading columns that are “not used a lot” while you don’t experience performance problems seems like premature optimization.
When dealing with really big columns (images, BLOBs…) though, you can use lazy loading, either through your ORM or by hand. The method that loads the column could be in the DAO, or in a specific service.