X-Git-Url: https://git.ralfj.de/rust-101.git/blobdiff_plain/188b1ec1b8528e2326791feccc8077e15bd60182..934a02211cd612a968195ded4bcc92e175dc0aff:/src/part13.rs?ds=sidebyside diff --git a/src/part13.rs b/src/part13.rs index 76d7154..8edcc99 100644 --- a/src/part13.rs +++ b/src/part13.rs @@ -7,10 +7,10 @@ use std::sync::mpsc::{sync_channel, SyncSender, Receiver}; use std::sync::Arc; //@ Our next stop are the concurrency features of Rust. We are going to write our own small version of "grep", -//@ called *rgrep*, and it is going to make use of concurrency: One thread reads the input files, one thread does +//@ called *rgrep*, and it is going to perform three jobs concurrently: One thread reads the input files, one thread does //@ the actual matching, and one thread writes the output. I already mentioned in the beginning of the course that //@ Rust's type system (more precisely, the discipline of ownership and borrowing) will help us to avoid a common -//@ pitfall of concurrent programming: data races. +//@ pitfall of concurrent programming: data races. We will see how that works concretely. // Before we come to the actual code, we define a data-structure `Options` to store all the information we need // to complete the job: Which files to work on, which pattern to look for, and how to output.
@@ -145,7 +145,7 @@ pub fn main() { run(options); } -// **Exercise 12.1**: Change rgrep such that it prints not only the matching lines, but also the name of the file +// **Exercise 13.1**: Change rgrep such that it prints not only the matching lines, but also the name of the file // and the number of the line in the file. You will have to change the type of the channels from `String` to something // that records this extra information. @@ -159,15 +159,15 @@ pub fn main() { //@ programs memory safe, and that prevents us from invalidating iterators, also helps secure our multi-threaded code against //@ data races. For example, notice how `read_files` sends a `String` to `filter_lines`. At run-time, only the pointer to //@ the character data will actually be moved around (just like when a `String` is passed to a function with full ownership). However, -//@ `read_files` has to *give up* ownership of the string to perform `send`, to it is impossible for an outstanding borrow to -//@ still be around. After it sent the string to the other side, `read_files` has no pointer into the string content +//@ `read_files` has to *give up* ownership of the string to perform `send`, to it is impossible for the string to still be borrowed. +//@ After it sent the string to the other side, `read_files` has no pointer into the string content //@ anymore, and hence no way to race on the data with someone else. //@ -//@ There is a little more to this. Remember the `'static` bound we had to add to `register` in the previous part, to make +//@ There is a little more to this. Remember the `'static` bound we had to add to `register` in the previous parts, to make //@ sure that the callbacks do not reference any pointers that might become invalid? This is just as crucial for spawning //@ a thread: In general, that thread could last for much longer than the current stack frame. Thus, it must not use //@ any pointers to data in that stack frame. This is achieved by requiring the `FnOnce` closure passed to `thread::spawn` -//@ to be valid for lifetime `'static`, as you can see in [its documentation](http://doc.rust-lang.org/stable/std/thread/fn.spawn.html). +//@ to be valid for lifetime `'static`, as you can see in [its documentation](https://doc.rust-lang.org/stable/std/thread/fn.spawn.html). //@ This avoids another kind of data race, where the thread's access races with the callee deallocating its stack frame. //@ It is only thanks to the concept of lifetimes that this can be expressed as part of the type of `spawn`. @@ -179,13 +179,14 @@ pub fn main() { //@ //@ The answer is already hinted at in the error: It will say something about `Send`. You may have noticed that the closure in //@ `thread::spawn` does not just have a `'static` bound, but also has to satisfy `Send`. `Send` is a trait, and just like `Copy`, -//@ it's just a marker - there are no functions provided by `Send`. What the trait says is that types which are `Send`, can be +//@ it's just a marker - there are no functions provided by `Send`. What the trait says is that types which are `Send` can be //@ safely sent to another thread without causing trouble. Of course, all the primitive data-types are `Send`. So is `Arc`, -//@ which is why Rust accepted our code. But `Rc` is not `Send`, and for a good reason! +//@ which is why Rust accepted our code. But `Rc` is not `Send`, and for a good reason! If had two `Rc` to the same data, and +//@ sent one of them to another thread, things could go havoc due to the lack of synchronization. //@ //@ Now, `Send` as a trait is fairly special. It has a so-called *default implementation*. This means that *every type* implements //@ `Send`, unless it opts out. Opting out is viral: If your type contains a type that opted out, then you don't have `Send`, either. //@ So if the environment of your closure contains an `Rc`, it won't be `Send`, preventing it from causing trouble. If however every //@ captured variable *is* `Send`, then so is the entire environment, and you are good. -//@ [index](main.html) | [previous](part12.html) | [next](part14.html) +//@ [index](main.html) | [previous](part12.html) | [raw source](workspace/src/part13.rs) | [next](part14.html)