Hello, fellow Django developers! I need an advice.
I have a typical Product model with descendants (Book, Laptop etc) that hold product-specific fields. Currently, I am using django-polymorphic, but I keep running into opinion that multi-table inheritance is a messy business and should be avoided if possible (e.g. this: https://jacobian.org/2010/nov/2/concrete-inheritance/). Besides that, I believe there is a rule that says something like “application level logic shouldn’t affect the way your data is stored”. And with multi-table inheritance it kinda does.
With that being said, I am not sure how to do this in a neat way, using abstract inheritance. Django-mptt to store our inheritance tree + contenttypes? Something along these lines?
I haven’t found packages that help to deal with abstract inheritance, even though it seems to be a common problem and a logical way to approach it. So maybe I am missing something?
To be honest, I am not sure if I should be concerned about it. I just keep coming back to it for some reason. Maybe I just need someone experienced in this problems to say some wise words that will calm my mind. Anyway, thank you in advance for answering.
<opinion>
This is a case of asking X number of people this question and getting 2X+3 number of opinions.
I agree that “application logic shouldn’t affect the way your data is stored” - however, that’s only half the picture.
The other half is that application logic requirements should affect the way your data is stored. Data models themselves don’t exist in a vacuum. There’s a set of processing requirements that drive the data model.
As you’ve already identified, there are a number of different ways to structure that type of data. Your optimal choices are going to depend upon what you need to do with it.
You mentioned the basic idea of a Product having descendants such as Book and Laptop. The next step then is to identify, within your system, what is the difference between a Book and a Laptop? For example, if you’re creating an inventory system, there may not be any type of effective difference at all. They’re just “things” where you’re tracking attributes such as location or ownership. Or, you might be building a system where you’re calculating power requirements for electrical outlets. In those cases, a Laptop has very real differences from Books that matter for this.
Bottom line from my perspective then is that there is no pat answer. You need to look beyond the entities themselves to identify the function those entities will perform. It’s that information that will best drive your data design.
</opinion>