3 ms·

If you are sophisticated enough to design a good relational model and use an ORM to generate it, what does an ORM give you? Serious question. The answers I've s

by dkarl 3y ago

If you are sophisticated enough to design a good relational model and use an ORM to generate it, what does an ORM give you? Serious question. The answers I've seen typically stress the advantages for developers who are uncomfortable with SQL (who I think are either going to get comfortable or fail, and the ORM will only delay them getting comfortable or enable them to create a more expensive failure) or more easily generate lots of different complex queries (which sounds plausible, but I've never seen it.)

mrweasel 3y ago
The ORM, in my mind, is just there to help you translate data into objects, that's it really. You could do that manually and I have worked on projects where we did just that, but it's a lot of repetitive work that brings little value. I have seen and done complex queries using a DSL for an ORM, in my case Django, but now you're just learning a different query language, so you're right that the ORM doesn't bring much to the table. Realistically, those who are uncomfortable with SQL are going to create poor queries with the ORM as well. For quick prototyping and systems with limited amounts of data ORMs can speed up development quite a bit. Technically there's a cost, but computers are fast enough that it doesn't matter on the small scale.
- dkarl 3y ago
  I guess the way you describe it is the way I like to work, keeping my object model in code close to the relational model to minimize the mental and performance cost of mapping back and forth, and using a SQL library to minimize boilerplate. I don't think of it as using ORM because my favorite tools for working that way don't bill themselves as ORMs, but I'm not sure what I'd use in Python other than SQLAlchemy. Even projects that seem to be stripped down non-ORMs like SQLModel turn out to be built on top of SQLAlchemy.
skydhash 3y ago
Before it was about “What if we need to change the database DBMS?” But now it’s more readable code and easy conversion to the language data structures. But it’s the first thing that is looked at when improving performance

by dkarl 3y ago

mrweasel 3y ago

The ORM, in my mind, is just there to help you translate data into objects, that's it really. You could do that manually and I have worked on projects where we did just that, but it's a lot of repetitive work that brings little value. I have seen and done complex queries using a DSL for an ORM, in my case Django, but now you're just learning a different query language, so you're right that the ORM doesn't bring much to the table. Realistically, those who are uncomfortable with SQL are going to create poor queries with the ORM as well. For quick prototyping and systems with limited amounts of data ORMs can speed up development quite a bit. Technically there's a cost, but computers are fast enough that it doesn't matter on the small scale.

dkarl 3y ago
I guess the way you describe it is the way I like to work, keeping my object model in code close to the relational model to minimize the mental and performance cost of mapping back and forth, and using a SQL library to minimize boilerplate. I don't think of it as using ORM because my favorite tools for working that way don't bill themselves as ORMs, but I'm not sure what I'd use in Python other than SQLAlchemy. Even projects that seem to be stripped down non-ORMs like SQLModel turn out to be built on top of SQLAlchemy.

skydhash 3y ago

Before it was about “What if we need to change the database DBMS?” But now it’s more readable code and easy conversion to the language data structures. But it’s the first thing that is looked at when improving performance