Hi NHibernate developers, I'm not filing this as a bug; the last time I tried that, it didn't go so well. ;-)
In order to get NHibernate to work with my legacy application, I had to comment out an optimization that was being made in order to try to prevent NHibernate from issuing an "update" statement rather than an "insert" statement if NHibernate was under the impression that the row in the joined table did not exist. My legacy application has a concept of join-table, but it doesn't remove the row if someone updates the value in it to be NULL like NHIbernate does. So you're left with a row with a primary key and no values in any of the other columns. Anyway, I'll let the test cases and code speak for itself, but feel free to use these test cases as you see fit. By now I've forked NHibernate for internal use within my company, but will still share patches and test cases as I come up with them. Regards, Mike
JoinTable.patch
Description: Binary data
