There are two possibilities: the first (and probably more probable) -
I'm using the modeler the wrong way and should listen much more than I
talk, and the second - modeler usage efficiency could be significantly
improved.
From what I can tell, the modeler (v1.2M1) doesn't handle reverse
engineering of a database over an existing ORM well: relations and/or
objects are duplicated, the validator reports loads of errors etc. My
routine so far was to basically delete everything I haven't modified
manually (which is usually almost everything), reengineer the database
and generate the source code and primary key support again. This is,
obviously, a tedious procedure.
It seems to me that the idea with two levels of generated classes could
be applied here, as well. The first level is always run over by database
reengineering, but the second one contains and protects modifications
users make on the object or database model.
I look forward to hearing your comments...
Tomislav
This archive was generated by hypermail 2.0.0 : Tue Apr 19 2005 - 09:42:55 EDT