|
From: Thomas R. <tri...@tr...> - 2003-04-16 16:30:38
|
Rod, I'm pretty much in agreement with your position. What initially attracted me to your JDBC framework was that it freed me from all the exception/connection closing issues. Another plus is that it lets you work in a more object oriented fashion. I don't mind adding layers on top of the basic Spring JDBC functionality, that would provide a comlete set of the insert/select/update/delete functionality in a nice and easy to use abstraction. For the primary key generation I like to use the functionality that the database provides and isolate it in a java class that can be swapped for a different implementation when you switch to a different database platform. What I like to do for Oracle is use a sequence, and have an insert-trigger that will look up the next value in the sequence if no primarky key value is supplied. Then I also have a database function that gets the next value from the sequence and returns it. This function can be used from any PL/SQL code that insert rows. On top of this I would have a "getNextKey" method to access this database function in a Java class that wraps any database specific code. Thomas > Isabelle, > > You make a valid point about not wanting to need another product for just > one problem. > > My view is that > - we don't want to get into full-blown O/R mapping > - BUT there is scope in building more JDBC-based functionality and > abstractions on top of the Spring JDBC way of doing things. To my mind the > central value proposition of Spring JDBC is the exception hierarchy and way > it frees developers of the tedious and error-prone tasks of raw JDBC. > Additional features can validly be layered on top. > > So I think you should propose how you think we should add the insert > functionality, perhaps even implement it, so we can discuss it and > incorporate it if we're all comfortable with it. I think it's > - important that any abstraction shouldn't impose potential inefficiency, > and should allow the use of efficient RDBMS functions like Oracle sequences > - important that it should be compatible with JDBC 3.0 moving forward. > > Thomas, do you have any thoughts on this? > > Regards, > Rod > > > Personally, I don't think I'd want to commit to yet another framework such > as hibernate for ex. in the project at work we're planning to do with > spring. Most of what we need for jdbc is in spring, except for the insert > functionality. Even with oracle, I don't like the idea of having to write a > stored procedure for every single insert, I'd probably opt for the > additional cost of a select next_val from dual in jdbc code (i.e. instead of > just doing an insert, first retrieve the key val then do the insert). The > main problem is with db's that don't have sequences, and if we agree to live > with holes in the primary key sequences we can implement an algorithm where > some sort of key manager per VM reserves a range of key values per table. > > > > Isabelle > > > > > > ------------------------------------------------------- > This sf.net email is sponsored by:ThinkGeek > Welcome to geek heaven. > http://thinkgeek.com/sf > _______________________________________________ > Springframework-developer mailing list > Spr...@li... > https://lists.sourceforge.net/lists/listinfo/springframework-developer > > |