►Opinion◄ Hierarchial vs. Composition Data Modeling.

BenHayat

Have you ever done a project in Aware where in cases where you needed to model the data as parent/child relation (hierarchical), but you chose to model it as reference, where no data owns another data?
In Parent/Child you basically create a hierarchical model, where as in reference relation, you build composition model.

If you did build the whole project based on composition modeling, what were your outcome?
Did it cause any problems?
Were queries faster, because Aware didn't do "Eager Loading"?
What were the benefits of not doing hierarchical modeling?

Thank you in advance!

BLOMASKY

I was under the impression that the only diff. in aware between Owner / Owned by and peer is that Owner performs a cascading delete if the owner is deleted. You mention Eagar loading in queries. Are you telling me that the sql query is different depending on the type of relationship? (I assume we are talking about a "to 1" relationship. If so, that would be good to know.

Frankly, I always thought that all relationships in Aware are Aware's implementation of SQL foreign keys. So my data modeling does not change if I implement Peer or Owner relationships. In SQL I handle the options by either making the relationship Identifying or Non-Identifying, and I have triggers to handle cascading changes / deletes.

Of course, I guess there is more than one way to handle modeling.

Bruce

BenHayat

BLOMASKY wrote
I was under the impression that the only diff. in aware between Owner / Owned by and peer is that Owner performs a cascading delete if the owner is deleted. You mention Eagar loading in queries. Are you telling me that the sql query is different depending on the type of relationship? (I assume we are talking about a "to 1" relationship. If so, that would be good to know.

Frankly, I always thought that all relationships in Aware are Aware's implementation of SQL foreign keys. So my data modeling does not change if I implement Peer or Owner relationships. In SQL I handle the options by either making the relationship Identifying or Non-Identifying, and I have triggers to handle cascading changes / deletes.

Of course, I guess there is more than one way to handle modeling.

Bruce

Bruce, I had made a mistake of using the wrong word in my original post. I had said "Peer" which I meant reference to another object. So, two objects have relationships as independent. My mistake!
Honestly, as I design classes and inheritance/composition in C# , I'm trying to see what's the closest formation I can do in Aware. That's where this idea came to mind. When a prospects says "Hey Ben, how would you model this (a C# collection of classes) in Aware?", and I want to see how close I can get to what they do in real OO model. Hope this clarifies my thoughts.