The thing with OOP, particularly how it’s used in GCed languages, is that it’s all about handing references out to wherever and then dealing with the complexity of not knowing who has access to your fields via getters & setters, or by cloning memory whenever it’s modified in asynchronous code.
Rust has quite the opposite mindset. It’s all about tracking where references go. It pushes your code to be very tree-shaped, i.e. references typically¹ only exist between a function and the functions it calls underneath. This is what allows asynchronous code to be safe in Rust, and I would also argue that the tree shape makes code easier to understand, too.
But yeah, some of the patterns you might know from OOP will not work in Rust for that reason. You will likely need to get into a different mindset over time.
Also just in case: We are talking OOP in the sense of the paradigm, i.e. object-oriented.
Just using objects, i.e. data with associated functions/methods, that works completely normal in Rust.
¹) If you genuinely need references that reach outside the tree shape, which is mostly going to be the case, if you work with multiple threads, then you can do so by wrapping your data structures in Arc<Mutex<_>> or similar. But yeah, when learning, you should try to solve your problems without these. Most programs don’t need them.
Hmm, not sure, if I’ve heard of it. I’m guessing, we’re not talking about simply drawing a UML class diagram…? Is it for figuring out which object will have to clean up which other objects, in non-GCed languages?
Yes, pretty much like UML diagrams. Who is responsible for allocating memory and freeing it.
Languages like Swift, Objective-C, C++ have features that mean you don’t need to do this by hand. But you have to tell the compiler if you want to keep and object around and who owns it.
Ah, interesting. I went from garbage-collected languages where thinking about ownership might be useful for keeping complexity low and occasionally comes up when you manage lists of objects, but ultimately isn’t needed, to Rust where well-defined ownership is enforced by the language.
So, I wasn’t aware that ownership is even as concrete of a thing in other languages…
Oh you need this in garbage collected languages too, once you run into memory use issues. GC languages are notorious for being wasteful with memory, even when working correctly.
Not if you never get your application into production…
I wish this was as much of a joke as I’m pretending. It’s so common for software projects to get cancelled that lots of tooling differences are just in terms of how long they let you not deal with long-term problems and how violently they do then explode into your face.
For most of the development lifecycle of a GCed project, you’re gonna ignore memory usage. And if you’re lucky, it can be ‘solved’ by just plonking down a thiccer piece of hardware…
This is basically where my learning took me. I had to develop this notion that there was a preferred directionality to ownership and data flow, like “grain” in a piece of wood. Everything is easier if you go with the grain. “Tree-shaped” works too, since it basically is the call graph of a (single threaded) program.
The point where I realized all this was when I tried to do a very Python/JS-brained thing: return a closure from a function. The moment you try to “curry” values into the closure, you have to “move” them to solve for ownership, lest you bring timelines into the picture. Which isn’t always what you want in a generic and reusable function. And sure enough, the standard lib and other popular libraries want you to pass a closure to functions instead.
The thing with OOP, particularly how it’s used in GCed languages, is that it’s all about handing references out to wherever and then dealing with the complexity of not knowing who has access to your fields via getters & setters, or by cloning memory whenever it’s modified in asynchronous code.
Rust has quite the opposite mindset. It’s all about tracking where references go. It pushes your code to be very tree-shaped, i.e. references typically¹ only exist between a function and the functions it calls underneath. This is what allows asynchronous code to be safe in Rust, and I would also argue that the tree shape makes code easier to understand, too.
But yeah, some of the patterns you might know from OOP will not work in Rust for that reason. You will likely need to get into a different mindset over time.
Also just in case: We are talking OOP in the sense of the paradigm, i.e. object-oriented.
Just using objects, i.e. data with associated functions/methods, that works completely normal in Rust.
¹) If you genuinely need references that reach outside the tree shape, which is mostly going to be the case, if you work with multiple threads, then you can do so by wrapping your data structures in
Arc<Mutex<_>>or similar. But yeah, when learning, you should try to solve your problems without these. Most programs don’t need them.OOP also has object ownership hierarchy structures. Which object owns which other object, is a question always worth answering.
Hmm, not sure, if I’ve heard of it. I’m guessing, we’re not talking about simply drawing a UML class diagram…? Is it for figuring out which object will have to clean up which other objects, in non-GCed languages?
Yes, pretty much like UML diagrams. Who is responsible for allocating memory and freeing it.
Languages like Swift, Objective-C, C++ have features that mean you don’t need to do this by hand. But you have to tell the compiler if you want to keep and object around and who owns it.
See this article on Objective-C to see the different ways to manage memory this language supports.
Ah, interesting. I went from garbage-collected languages where thinking about ownership might be useful for keeping complexity low and occasionally comes up when you manage lists of objects, but ultimately isn’t needed, to Rust where well-defined ownership is enforced by the language.
So, I wasn’t aware that ownership is even as concrete of a thing in other languages…
Oh you need this in garbage collected languages too, once you run into memory use issues. GC languages are notorious for being wasteful with memory, even when working correctly.
Not if you never get your application into production…
I wish this was as much of a joke as I’m pretending. It’s so common for software projects to get cancelled that lots of tooling differences are just in terms of how long they let you not deal with long-term problems and how violently they do then explode into your face.
For most of the development lifecycle of a GCed project, you’re gonna ignore memory usage. And if you’re lucky, it can be ‘solved’ by just plonking down a thiccer piece of hardware…
Great points. Especially putting in more memory can get you very far.
Database optimization? Nah, just put in 1 TB of RAM to keep the whole DB in memory at all times.
This is basically where my learning took me. I had to develop this notion that there was a preferred directionality to ownership and data flow, like “grain” in a piece of wood. Everything is easier if you go with the grain. “Tree-shaped” works too, since it basically is the call graph of a (single threaded) program.
The point where I realized all this was when I tried to do a very Python/JS-brained thing: return a closure from a function. The moment you try to “curry” values into the closure, you have to “move” them to solve for ownership, lest you bring timelines into the picture. Which isn’t always what you want in a generic and reusable function. And sure enough, the standard lib and other popular libraries want you to pass a closure to functions instead.