I'm doing research if hibernate could be usefull for future projects as a "ligth" version of EJB CMP. I'm using JBuilder as my primary IDE and liked modelling relations between EJB's with the built-in modeller/gui.

My main question is: is there also such a gui to model relations and field/method's for hibernate?

I've done some research and i suspect i've found a way:
- poseidon:
http://www.gentleware.com/products/poseidonCE.php3
To model the relations and fields/methodes. And export model to .xmi format.

- andromda:
http://www.andromda.org/index.html
To convert the .xmi from poseidon to .hbm format

- hibernate tools:
To convert the .hbm to .ddl and/or .java

Could this work and are there other ways to convert uml model to .hbm format? Suggestions on maybe better uml tools which support xmi are also welcome.

I'm not aware of any other way and I also heard people had some success with the process you describe. But I'm also sure you don't get full roundtrip development including the UML diagrams. I doubt there is a way in a heterogeneous system which fulfils all the functions with all the extras required for more than prototype work.

Thanks for the quick answer! I'm gonna try the steps from my orginal post and see if there are any problems and how big these problems are.

I have done a fair amount of work, including overcoming some bugs and other obstacles, to get AndroMDA generating code for Hibernate. I have two big issues:

1. The cartridge seems to only generate EJB code. I was hoping to get POJOs or even Struts. I was wondering if there are other flavors of the Hibernate cartridge that are not EJBs.

2. Is there any documentation on the cartridge? I have had to sift through the Velocity templates to discover some of the required stereotypes that make this cartridge work, but it's a lot to absorb for someone who is new to UML, Velocity, Hibernate, and Xdoclet.

I know that I need Entity's and Service's, but I'm not clear on the difference between the two. I see that each class must have a PrimaryKey, but I'm not sure how to deal with compound primary keys. I see a number of other stereotypes mentioned in the templates, but I am not at all clear on what their significance is. Is there any documentation at all on this stuff?

Thanks,
Lee Grey
Grey Matter

We made a tool named Anything from Xmi GENerator (AXgen) available at http://axgen.sourceforge.net. We originally designed it to generate repository, entity classes and db_scheme.dll for use with OJB.

AXgen uses velocitiy templates to transform XMI to anything you want. The current release includes only templates for OJB, but we are working on templates for use with hibernate but as we have some architectural problems, we are a little stamped. But I could publish my drafts, if somebody is interested in. You could also try to write you're own templates. It's not that hard. Maybe you want to give it a try.

Peter

I'm just getting started with AndroMDA on Hibernate as well. I suspect I am opening a pandora's box (like... "oh *#$* it doesn't do x? This thing is due in the morning!!), but armed with the sources and the will to make it work, it should be a lot of fun.

The thing I keep telling myself, and warrants remembering, is how little you had available to you under CMP. So when I hit a bottleneck, I first ask myself whether I could have done it in CMP, and if not, take my lumps. If so, I ask myself how long it would have taken. If it would have taken a long time, but I really need it done that bad, I consider whether the time I would have otherwise spent on the broken methods of the past could be invested in helping the tool along. I suspect that to be the case here.

Take a look at the source. It's really not that complicated. Once you get past that initial "here we go again" feeling with opening the trunk on another open source tool, it's pretty easy to see where to go next. Truth be told though, the last time I looked at AndroMDA was early August. I think my brain has settled a bit since then on how/when/why to use it.

I agree with you that AndroMDA has a ways to go to cover all the cases that one can express in UML, but in the 80/20 view, the 80% is there, IMHO. It's up to us now to start making the 20% happen.

<insert_standard_disclaimers/>

-b