Comments on: ORM-is-Dead meme

By: Blog harvest, October II « Schneide Blog

Blog harvest, October II « Schneide Blog — Sun, 25 Oct 2009 19:21:55 +0000

[...] how to use it – and when not to. Replies followed instantly, here are two noteworthy ones by Scot Mcphee and by Jens [...]

By: Dan Howard

Dan Howard — Tue, 20 Oct 2009 13:56:20 +0000

Why do people keep torturing themselves with Hibernate anyway?

http://code.google.com/p/persist/

Wood simple and is never in your way.

By: uberVU - social comments

uberVU - social comments — Tue, 20 Oct 2009 01:49:35 +0000

Social comments and analytics for this post…

This post was mentioned on Twitter by scotartt: ORM-is-Dead meme http://ff.im/-a6dyu...

By: Tweets that mention let x=x › ORM-is-Dead meme -- Topsy.com

Tweets that mention let x=x › ORM-is-Dead meme -- Topsy.com — Mon, 19 Oct 2009 18:12:11 +0000

[...] This post was mentioned on Twitter by Stephan Schmidt and scot mcphee, Willy Mejía. Willy Mejía said: RT @codemonkeyism: RT @scotartt: ORM-is-Dead meme http://ff.im/-a6dyu [...]

By: Robert Watkins

Robert Watkins — Mon, 19 Oct 2009 12:02:30 +0000

*shrug* If it’s transparent to the next tier, that means they won’t know if you put it in, or if you take it out. So why complicate things before it’s necessary?

Hibernate can do that, if you put in a repository pattern to handle loads & searches, etc. I agree totally that your web app shouldn’t need to import org.hibernate (or iBatis, or anything) – that’s just silly.

My domain model trumps the database design – if it’s easy to map, I do it directly, if it’s not easy to map, well that’s where things like components & custom types come in.

But my strong preference isn’t to have a database at all. ORM isn’t dead, but it is dead boring…

By: Scot Mcphee

Scot Mcphee — Mon, 19 Oct 2009 11:32:43 +0000

You know my position on this Robert … why would you ties your interface domain model to your database design? If it’s that simple, you barely need an ORM anyway!

By: Robert Watkins

Robert Watkins — Mon, 19 Oct 2009 10:07:58 +0000

The need to have annotations in your classpath to load classes was an early Java 5 bug – more recent versions of Java 5, and ALL versions of Java 6, do not need annotations present.

So, for example, if you’re building a classic EAR with EJB and WAR, or an OSGi-app with good classloader partioning, you can have a DAO that is full of hibernate annotations and not have said annotations on the classpath of the web app/client app/etc.

Of course, if you’re using Maven to do this, it’s more painful – Maven insists on pushing compile-scope out to end users.

By: Stephan Schmidt

Stephan Schmidt — Mon, 19 Oct 2009 08:30:13 +0000

“[...] you’ve just exploded the whole point of layering systems right there!”

Exactly.

By: Scot Mcphee

Scot Mcphee — Mon, 19 Oct 2009 08:24:25 +0000

Oh … I find Hibernate annotations evil. Generally if I find teams really wanting to use them I then ask that the objects so annotated never go past the DAO layer, i.e. do not form part of the DAO signature. You should see the consternation that causes! The idea that some web command object using a service layer has to import org.hibernate.anything is beyond comprehension — you’ve just exploded the whole point of layering systems right there!

There’s an additional wrongness about hibernate annotations (and in fact most uses of annotations that I see). Whatever happened to the separation of code from configuration? I don’t find any problem with using the .hbm.xml files. People who say “it makes you look in two files” are wrong – you just need the .hbm.xml file to grok the mapping.

By: Stephan Schmidt

Stephan Schmidt — Mon, 19 Oct 2009 08:05:24 +0000

“Also, I get a bit annoyed about ORM abstractions leaking into the web tier, this is especially prevalent with using annotations rather than ORM mapping files.”

Hurray, another soul who doesn’t think annotations are the way to go with Hibernate.

Cheers
Stephan
http://www.codemonkeyism.com