On Feb 8, 2010, at 3:39 PM, Kevin Menard wrote:
>
> I'm reaching here a bit since it's been a while since I looked at this
> problem, but I think the semantics were preserved because
> isSubentityOf
> wasn't a proper subentity relationship. I.e., an entity could be a
> subentity of itself. So, while the code looks strange, it worked.
- if (rel.getTargetEntity() != src)
+ if (target.isSubentityOf((ObjEntity)
rel.getTargetEntity())) {
continue;
+ }
In the past we'd give up on a given relationship as a potential
candidate for a reverse relationship based on the fact that its
*target* entity is not the same as forward relationship *source*.
The new code compares inheritance hierarchy of the forward
relationship *target*, and reverse candidate relationship *target*.
Apples and oranges. Seems like the "if" check would *always* fail here
except for self-referential relationships? So the check itself is not
doing anything, and simply lets the following code to execute
unconditionally.
And the result is that further analysis is done based on the
DbRelationship path, and relationships starting/terminating at
different subentities of the same inheritance hierarchy as picked as a
forward/reverse pair (which IMO is incorrect until we implement a more
sophisticated matching algorithm as discussed).
So based on that conclusion I am going to revert the CAY-1009 patch
(and also commit CAY-1378 tests).
Andrus
This archive was generated by hypermail 2.0.0 : Wed Feb 10 2010 - 15:51:10 EST