diff options
Diffstat (limited to 'src/doc/guide-pointers.md')
| -rw-r--r-- | src/doc/guide-pointers.md | 492 |
1 files changed, 492 insertions, 0 deletions
diff --git a/src/doc/guide-pointers.md b/src/doc/guide-pointers.md new file mode 100644 index 00000000000..19696b42a37 --- /dev/null +++ b/src/doc/guide-pointers.md @@ -0,0 +1,492 @@ +% The Rust Pointer Guide + +Rust's pointers are one of its more unique and compelling features. Pointers +are also one of the more confusing topics for newcomers to Rust. They can also +be confusing for people coming from other languages that support pointers, such +as C++. This guide will help you understand this important topic. + +# You don't actually need pointers + +I have good news for you: you probably don't need to care about pointers, +especially as you're getting started. Think of it this way: Rust is a language +that emphasizes safety. Pointers, as the joke goes, are very pointy: it's easy +to accidentally stab yourself. Therefore, Rust is made in a way such that you +don't need them very often. + +"But guide!" you may cry. "My co-worker wrote a function that looks like this: + +~~~rust +fn succ(x: &int) -> int { *x + 1 } +~~~ + +So I wrote this code to try it out: + +~~~rust{.ignore} +fn main() { + let number = 5; + let succ_number = succ(number); + println!("{}", succ_number); +} +~~~ + +And now I get an error: + +~~~ {.notrust} +error: mismatched types: expected `&int` but found `<VI0>` (expected &-ptr but found integral variable) +~~~ + +What gives? It needs a pointer! Therefore I have to use pointers!" + +Turns out, you don't. All you need is a reference. Try this on for size: + +~~~rust +# fn succ(x: &int) -> int { *x + 1 } +fn main() { + let number = 5; + let succ_number = succ(&number); + println!("{}", succ_number); +} +~~~ + +It's that easy! One extra little `&` there. This code will run, and print `6`. + +That's all you need to know. Your co-worker could have written the function +like this: + +~~~rust +fn succ(x: int) -> int { x + 1 } + +fn main() { + let number = 5; + let succ_number = succ(number); + println!("{}", succ_number); +} +~~~ + +No pointers even needed. Then again, this is a simple example. I assume that +your real-world `succ` function is more complicated, and maybe your co-worker +had a good reason for `x` to be a pointer of some kind. In that case, references +are your best friend. Don't worry about it, life is too short. + +However. + +Here are the use-cases for pointers. I've prefixed them with the name of the +pointer that satisfies that use-case: + +1. Owned: ~Trait must be a pointer, because you don't know the size of the +object, so indirection is mandatory. +2. Owned: You need a recursive data structure. These can be infinite sized, so +indirection is mandatory. +3. Owned: A very, very, very rare situation in which you have a *huge* chunk of +data that you wish to pass to many methods. Passing a pointer will make this +more efficient. If you're coming from another language where this technique is +common, such as C++, please read "A note..." below. +4. Managed: Having only a single owner to a piece of data would be inconvenient +or impossible. This is only often useful when a program is very large or very +complicated. Using a managed pointer will activate Rust's garbage collection +mechanism. +5. Reference: You're writing a function, and you need a pointer, but you don't +care about its ownership. If you make the argument a reference, callers +can send in whatever kind they want. + +Five exceptions. That's it. Otherwise, you shouldn't need them. Be sceptical +of pointers in Rust: use them for a deliberate purpose, not just to make the +compiler happy. + +## A note for those proficient in pointers + +If you're coming to Rust from a language like C or C++, you may be used to +passing things by reference, or passing things by pointer. In some languages, +like Java, you can't even have objects without a pointer to them. Therefore, if +you were writing this Rust code: + +~~~rust +# fn transform(p: Point) -> Point { p } +struct Point { + x: int, + y: int, +} + +fn main() { + let p0 = Point { x: 5, y: 10}; + let p1 = transform(p0); + println!("{:?}", p1); +} + +~~~ + +I think you'd implement `transform` like this: + +~~~rust +# struct Point { +# x: int, +# y: int, +# } +# let p0 = Point { x: 5, y: 10}; +fn transform(p: &Point) -> Point { + Point { x: p.x + 1, y: p.y + 1} +} + +// and change this: +let p1 = transform(&p0); +~~~ + +This does work, but you don't need to create those references! The better way to write this is simply: + +~~~rust +struct Point { + x: int, + y: int, +} + +fn transform(p: Point) -> Point { + Point { x: p.x + 1, y: p.y + 1} +} + +fn main() { + let p0 = Point { x: 5, y: 10}; + let p1 = transform(p0); + println!("{:?}", p1); +} +~~~ + +But won't this be inefficient? Well, that's a complicated question, but it's +important to know that Rust, like C and C++, store aggregate data types +'unboxed,' whereas languages like Java and Ruby store these types as 'boxed.' +For smaller structs, this way will be more efficient. For larger ones, it may +be less so. But don't reach for that pointer until you must! Make sure that the +struct is large enough by performing some tests before you add in the +complexity of pointers. + +# Owned Pointers + +Owned pointers are the conceptually simplest kind of pointer in Rust. A rough +approximation of owned pointers follows: + +1. Only one owned pointer may exist to a particular place in memory. It may be +borrowed from that owner, however. +2. The Rust compiler uses static analysis to determine where the pointer is in +scope, and handles allocating and de-allocating that memory. Owned pointers are +not garbage collected. + +These two properties make for three use cases. + +## References to Traits + +Traits must be referenced through a pointer, because the struct that implements +the trait may be a different size than a different struct that implements the +trait. Therefore, unboxed traits don't make any sense, and aren't allowed. + +## Recursive Data Structures + +Sometimes, you need a recursive data structure. The simplest is known as a 'cons list': + +~~~rust +enum List<T> { + Nil, + Cons(T, ~List<T>), +} + +fn main() { + let list: List<int> = Cons(1, ~Cons(2, ~Cons(3, ~Nil))); + println!("{:?}", list); +} +~~~ + +This prints: + +~~~ {.notrust} +Cons(1, ~Cons(2, ~Cons(3, ~Nil))) +~~~ + +The inner lists _must_ be an owned pointer, because we can't know how many +elements are in the list. Without knowing the length, we don't know the size, +and therefore require the indirection that pointers offer. + +## Efficiency + +This should almost never be a concern, but because creating an owned pointer +boxes its value, it therefore makes referring to the value the size of the box. +This may make passing an owned pointer to a function less expensive than +passing the value itself. Don't worry yourself with this case until you've +proved that it's an issue through benchmarks. + +For example, this will work: + +~~~rust +struct Point { + x: int, + y: int, +} + +fn main() { + let a = Point { x: 10, y: 20 }; + spawn(proc() { + println!("{}", a.x); + }); +} +~~~ + +This struct is tiny, so it's fine. If `Point` were large, this would be more +efficient: + +~~~rust +struct Point { + x: int, + y: int, +} + +fn main() { + let a = ~Point { x: 10, y: 20 }; + spawn(proc() { + println!("{}", a.x); + }); +} +~~~ + +Now it'll be copying a pointer-sized chunk of memory rather than the whole +struct. + +# Managed Pointers + +> **Note**: the `@` form of managed pointers is deprecated and behind a +> feature gate (it requires a `#[feature(managed_pointers)];` attribute on +> the crate root; remember the semicolon!). There are replacements, currently +> there is `std::rc::Rc` and `std::gc::Gc` for shared ownership via reference +> counting and garbage collection respectively. + +Managed pointers, notated by an `@`, are used when having a single owner for +some data isn't convenient or possible. This generally happens when your +program is very large and complicated. + +For example, let's say you're using an owned pointer, and you want to do this: + +~~~rust{.ignore} +struct Point { + x: int, + y: int, +} + +fn main() { + let a = ~Point { x: 10, y: 20 }; + let b = a; + println!("{}", b.x); + println!("{}", a.x); +} +~~~ + +You'll get this error: + +~~~ {.notrust} +test.rs:10:20: 10:21 error: use of moved value: `a` +test.rs:10 println!("{}", a.x); + ^ +note: in expansion of format_args! +<std-macros>:158:27: 158:81 note: expansion site +<std-macros>:157:5: 159:6 note: in expansion of println! +test.rs:10:5: 10:25 note: expansion site +test.rs:8:9: 8:10 note: `a` moved here because it has type `~Point`, which is moved by default (use `ref` to override) +test.rs:8 let b = a; + ^ +~~~ + +As the message says, owned pointers only allow for one owner at a time. When you assign `a` to `b`, `a` becomes invalid. Change your code to this, however: + +~~~rust +struct Point { + x: int, + y: int, +} + +fn main() { + let a = @Point { x: 10, y: 20 }; + let b = a; + println!("{}", b.x); + println!("{}", a.x); +} +~~~ + +And it works: + +~~~ {.notrust} +10 +10 +~~~ + +So why not just use managed pointers everywhere? There are two big drawbacks to +managed pointers: + +1. They activate Rust's garbage collector. Other pointer types don't share this +drawback. +2. You cannot pass this data to another task. Shared ownership across +concurrency boundaries is the source of endless pain in other languages, so +Rust does not let you do this. + +# References + +References are the third major kind of pointer Rust supports. They are +simultaneously the simplest and the most complicated kind. Let me explain: +references are considered 'borrowed' because they claim no ownership over the +data they're pointing to. They're just borrowing it for a while. So in that +sense, they're simple: just keep whatever ownership the data already has. For +example: + +~~~rust +use std::num::sqrt; + +struct Point { + x: f32, + y: f32, +} + +fn compute_distance(p1: &Point, p2: &Point) -> f32 { + let x_d = p1.x - p2.x; + let y_d = p1.y - p2.y; + + sqrt(x_d * x_d + y_d * y_d) +} + +fn main() { + let origin = @Point { x: 0.0, y: 0.0 }; + let p1 = ~Point { x: 5.0, y: 3.0 }; + + println!("{:?}", compute_distance(origin, p1)); +} +~~~ + +This prints `5.83095189`. You can see that the `compute_distance` function +takes in two references, but we give it a managed and unique pointer. Of +course, if this were a real program, we wouldn't have any of these pointers, +they're just there to demonstrate the concepts. + +So how is this hard? Well, because we're ignoring ownership, the compiler needs +to take great care to make sure that everything is safe. Despite their complete +safety, a reference's representation at runtime is the same as that of +an ordinary pointer in a C program. They introduce zero overhead. The compiler +does all safety checks at compile time. + +This theory is called 'region pointers,' and involve a concept called +'lifetimes'. Here's the simple explanation: would you expect this code to +compile? + +~~~rust{.ignore} +fn main() { + println!("{}", x); + let x = 5; +} +~~~ + +Probably not. That's because you know that the name `x` is valid from where +it's declared to when it goes out of scope. In this case, that's the end of +the `main` function. So you know this code will cause an error. We call this +duration a 'lifetime'. Let's try a more complex example: + +~~~rust +fn main() { + let mut x = ~5; + if *x < 10 { + let y = &x; + println!("Oh no: {:?}", y); + return; + } + *x -= 1; + println!("Oh no: {:?}", x); +} +~~~ + +Here, we're borrowing a pointer to `x` inside of the `if`. The compiler, however, +is able to determine that that pointer will go out of scope without `x` being +mutated, and therefore, lets us pass. This wouldn't work: + +~~~rust{.ignore} +fn main() { + let mut x = ~5; + if *x < 10 { + let y = &x; + *x -= 1; + + println!("Oh no: {:?}", y); + return; + } + *x -= 1; + println!("Oh no: {:?}", x); +} +~~~ + +It gives this error: + +~~~ {.notrust} +test.rs:5:8: 5:10 error: cannot assign to `*x` because it is borrowed +test.rs:5 *x -= 1; + ^~ +test.rs:4:16: 4:18 note: borrow of `*x` occurs here +test.rs:4 let y = &x; + ^~ +~~~ + +As you might guess, this kind of analysis is complex for a human, and therefore +hard for a computer, too! There is an entire [guide devoted to references +and lifetimes](guide-lifetimes.html) that goes into lifetimes in +great detail, so if you want the full details, check that out. + +# Returning Pointers + +We've talked a lot about functions that accept various kinds of pointers, but +what about returning them? Here's the rule of thumb: only return a unique or +managed pointer if you were given one in the first place. + +What does that mean? Don't do this: + +~~~rust +fn foo(x: ~int) -> ~int { + return ~*x; +} + +fn main() { + let x = ~5; + let y = foo(x); +} +~~~ + +Do this: + +~~~rust +fn foo(x: ~int) -> int { + return *x; +} + +fn main() { + let x = ~5; + let y = ~foo(x); +} +~~~ + +This gives you flexibility, without sacrificing performance. For example, this will +also work: + +~~~rust +fn foo(x: ~int) -> int { + return *x; +} + +fn main() { + let x = ~5; + let y = @foo(x); +} +~~~ + +You may think that this gives us terrible performance: return a value and then +immediately box it up?!?! Isn't that the worst of both worlds? Rust is smarter +than that. There is no copy in this code. `main` allocates enough room for the +`@int`, passes a pointer to that memory into `foo` as `x`, and then `foo` writes +the value straight into that pointer. This writes the return value directly into +the allocated box. + +This is important enough that it bears repeating: pointers are not for optimizing +returning values from your code. Allow the caller to choose how they want to +use your output. + + +# Related Resources + +* [Lifetimes guide](guide-lifetimes.html) |