Hello,
I am using Hibernate Tools 3.1.0 alpha 5 in Eclipse 3.1.0 and have successfully reverse engineered my db into mapping files and pojos. (very cool stuff)

However, I am forced to edit the mapping files slightly because:
1: they conatain <generator class="assign" > for fields that are 'serial' primary keys (PostgreSQL 8.0). I would like them to be "identity"

2: for the 'one-to-many' associations, there is no 'cascade' attribute in the <set> element. For hibernate to correctly save the set of associated types, I am adding cascade="all-delete-orphan"

My question is: Is there a way to configure hibernate so the hbm.xml files that are created do not have to be edited?

Thanks in advance for any suggestions.

Check out the stuff on reverse engineering http://www.hibernate.org/hib_docs/tools/ant/index.html#gen10 . If you don't find it there, you have just found a great reason to use XSLT ;-)

Max, if H Tools had a sort of second-pass-transformation feature, do you think anyone would use it?

well we have it implicitly today (jtidy runs on the xml), so if we could make the general available (need to collect the generated files etc.) it might be ok...but don't know if it really is going to be used...

btw. this is solved now in the current cvs version - here you can specify the generator and parameters per table.

Would I specify these in the hibernate.reveng.xml file? Could you point me to an example of the configuration?

Thanks!

currently only in cvs - there is some junit tests with reveng.xml