On Aug 5, 2008, at 9:35 AM, Marcin Skladaniec wrote:
>> I think you are talking about VERTICAL inheritance, not horizontal
>> here. In case of horizontal you wouldn't have a relationship
>> between subclass/superclass tables. Anyways, assuming we are
>> discussing vertical inheritance, yes, there is a relationship
>> involved. Per JPA approach, this relationship can be figured
>> automatically : it is a 1..1 relationship between the PK's of two
>> tables. So we don't have to store it in XML, as we can find it in
>> runtime. Still internally there should be a concept of a
>> relationship joining parent and child tables, so you are right
>> about the big picture.
>>
> I'm now confused with vertical-horizontal naming. You say vertical,
> jira https://issues.apache.org/cayenne/browse/CAY-795says
> horizontal. I guess we talk about the same thing anyway, but just to
> be sure: I'm trying to model a database schema where there are 3
> tables: Artwork, Painting and Sculpture.
> Artwork is an abstract entity with its own table
> Painting and Sculpture are extending Artwork, but have their own
> tables as well.
> You are suggesting that both superclass and subclass should have PK,
> this means yet another table in required to link the two tables.
> This seems odd to me as the relationship is simple 1..1, is that a
> requirement of JPA ?
> I'm reading up on JPA, but I cannot find it.
I guess CAY-795 is just not specific enough. It says "horizontal" in
the title, but the steps to implement it are described in a rather
generic fashion. Not sure if we have a separate Jira for vertical?
Maybe Ari can open a new Jira on that.
Thanks,
Andrus
This archive was generated by hypermail 2.0.0 : Tue Aug 05 2008 - 10:07:26 EDT