I introduced this patch, and admittedly did not fully consider the breaking
change that's been brought up. I would agree that the changes in the Sql2005
dialect should be removed and left up to the user. The pain that I was
trying to address was exactly what Davy mentioned; that as it was, NH
generated a schema that had clustered primary keys, which prevented me from
being able to use <database-object> to specify which columns I wanted to
cluster on. Once I had a means to change this behavior, my problem was
solved, so taking the change into the Sql2005 dialect without discussion was
a mistake on my part.

Reply via email to