- MD5 of json content stored on EntityBeanIntercept for dirty detection
- Only convert to JSON once (at dirty detection time). Store this json content on EntityBeanIntercept to later push to ScalarTypeJsonObjectMapper for bind
Should consider alternative to extend BeanProperty rather than have these if blocks.
- Changes ModifyHolder to use IdentityHashMap such that equals/hashcode isn't invoked
- Change SaveManyBeans such that M2M intersection deletes execute before inserts
The SaveManyBeans change is required to pass TestM2MModifyTest. With that test we get a delete/insert pair that we somewhat expect because we clear() the collection. Previously equals/hashCode meant we didn't get that delete/insert pair.
The test changes from containsExactly() to containsOnly() as we now can't guarantee the ordering with IdentityHashMap.
default method
Iterator.remove() has a default implementation that throws UnsupportedOperationException which is what we want. Remove the close() method from QueryIterator and implementations so that we just use the Iterator.remove() default implementation.
When the value is null and not changed then is was null before and should not be treated as dirty. Hence only check for dirty when value != null in BeanDescriptor 3182
Adds the ability to change the default "jsonDirtyByDefault" configuration setting used with DbJson Jackson properties. Currently these default to being assumed dirty and this allows us to change that to be assumed not dirty (which is very likely the better default).
Noting that there might be a better fix. The addJoin() at line 151 is suppressed as it is a duplicate for this case of a !oneToOneExported so a better fix might be to only add the soft delete clause when the join as actually added (not suppressed as a duplicate).
- Clears the buffer after it has been used to load BeanCollection's
- Sets the BeanCollection to use the server rather than the Load Buffer just in case there are subsequent calls for loading (which there should not be)
That is, currently a filterMany automatically marks that path as a fetchQuery path. This change turns that off and the filterMany path and predicates can then be included in the main query.
This is better for performance (to include the filterMany path in with the main query) when the query is a findEach and looking to return a large number of results.
rbygrave