Unit test for issue #1766

[fubar-coder]

This is a simplified version of NHibernate.Test.ListIndex.ListIndexFixture that uses an inverse list.

This unit test shows that an inverse list causes missing update statements to set the index column.

[fredericDelaporte]

The index has to be set here, like the parent. When a list is mapped as inverse, no update occurs from it. It may at most cascade the update/save operation if cascade is enabled, but it will not set the appropriate parent and index.

From here:

Changes made only to the inverse end of the association are not persisted.

The documentation also implies it with:

If there is a property of the child class that maps to the index column you can use inverse="true" on the collection mapping
...
If there is no such property on the child class, the association cannot be considered truly bidirectional. That is, there is information available at one end of the association that is not available at the other end. In this case, you cannot map the collection inverse="true".

This is so because once the collection is mapped as inverse, the index must be handled from the other side.

[fubar-coder]

Ah, ok. So it was a misunderstanding on my side. Would it be possible to update/clarify the documentation for this use case and/or write a unit test that shows this behavior? I did expect the index column to always be set/updated, because it's an integral part of the list.

[fredericDelaporte]

There is also:

The non-inverse side is used to save the in-memory representation to the database.

Anyway, a consequent update is also ongoing, awaiting review (see #1761). It notably adds some more examples and guidance, including some about lists.

I do not think we need to add more.