Just learning. I threw together a little CRUD API in Rocket the other day.
Now I’m playing around with Diesel. I don’t love the intermediate New types, coming from EF Core. Is that because Rust ~~~~doesn’t really have constructors?
Ah, I see. So you’re expecting to have one object for creation, updates, queries, etc.
I work with something like that at work (SQLAlchemy in Python), and I honestly prefer the Diesel design. I build an object for exactly what I need, so I’ll have a handful of related types used for different purposes. In Python, we have a lot of “contains_eager” calls to ensure data isn’t lazy loaded, and it really clutters up the code. With Diesel, that’s not necessary because I just don’t include data that I don’t need for that operation. Then again, I’m not generally a fan of ORMs and prefer to write SQL, so that’s where I’m coming from.
One of my main concerns with this is the potential for making a lot of separate calls to the DB for a complex data structure. Then again there are trade offs to any architecture.
Isn’t the reverse true? If you make separate models for each query, the ORM knows exactly what data you need, so it can fetch it all as once. If you use generic models, the ORM needs to guess, and many revert to lazy loading if it’s not sure (i.e. lots of queries).
That’s at least my experience with SQLAlchemy, we put a lot of effort in to reduce those extra calls because we’re using complex, generalized structures.
Just learning. I threw together a little CRUD API in Rocket the other day.
Now I’m playing around with Diesel. I don’t love the intermediate New types, coming from EF Core. Is that because Rust ~~~~doesn’t really have constructors?
Can you give an example? Pretty much everything in Diesel is abstracted away through trait macros.
The insert on their Getting Started guide.
Of course the other possibility is this guide is very low on abstractions.
Ah, I see. So you’re expecting to have one object for creation, updates, queries, etc.
I work with something like that at work (SQLAlchemy in Python), and I honestly prefer the Diesel design. I build an object for exactly what I need, so I’ll have a handful of related types used for different purposes. In Python, we have a lot of “contains_eager” calls to ensure data isn’t lazy loaded, and it really clutters up the code. With Diesel, that’s not necessary because I just don’t include data that I don’t need for that operation. Then again, I’m not generally a fan of ORMs and prefer to write SQL, so that’s where I’m coming from.
One of my main concerns with this is the potential for making a lot of separate calls to the DB for a complex data structure. Then again there are trade offs to any architecture.
Isn’t the reverse true? If you make separate models for each query, the ORM knows exactly what data you need, so it can fetch it all as once. If you use generic models, the ORM needs to guess, and many revert to lazy loading if it’s not sure (i.e. lots of queries).
That’s at least my experience with SQLAlchemy, we put a lot of effort in to reduce those extra calls because we’re using complex, generalized structures.