Test for incorrect audit table insert on tables with tree structure

[wojtek-abakus]

As requested in Issue #18 I have made a test for this example.

[RogerKratz]

Isn't the mapping a bit strange? Which side of the relationship should be responsible for the parentKey, Parent or Children? In your example it's both of them...

Either let Children be responsible for the relationship by setting update="false" insert="false" on Parent mapping, or map Children with inverse="true" (then you have to map index property as well).

One could say that because nhib supports this "both side mapping", envers should as well. Doesn't feel like normal mapping though.

[wojtek-abakus]

Yes you are right, in my original example using fluentnhibernate the collection was mapped as "inverse". I forgot to add this in the xml mapping file.

This doesn't change the test result though

[RogerKratz]

I didn't even know this mapping was allowed in NH Core? In your test, you have a bidirectional ref where "many-side" is marked with inverse=true - still that side is responsible for keeping track of the index position of the element? I was expecting to see a index mapped as a property on "one-side" to keep track of the index position if "many-side" was mapped with inverse=true (that would have worked in envers, I'm pretty sure without have tried it in your case).

[wojtek-abakus]

Hello, any chance for a fix of this issue ? Or could you be so kind and point me some workaround ?