Fighting the Borrow Checker
m-decoster.github.io
m-decoster.github.io
Is the borrow checker painful to grasp at first? A little. But the safety and assurance it gives you is immeasurable. Write enough Rust, and before long, you'll know when your code won't compile as written.
As a systems language, Rust _must_ give you total control when you need it. It's just not the default.
If you really don't want a hassle, then maybe you want to program in a garbage-collected language instead. But then you're potentially giving up some performance and/or memory usage savings.
struct Foo {
s1: String, // some sort of String data
s2: &str, // a slice of said String
}
In general, these cases are small, but they can be annoying when you run into them. As I said downthread, I very rarely run into them personally, but some others seem to constantly. YMMV.
What happens if you have ownership, then you can mutate the struct Foo like this:
foo.s1 = "World"
You couldn't because it'd be immutably borrowed.
Plenty of people have fought the type system in Java. Happens all the time with Generics and Type eraser. How many projects have work around for that Scala, Flink, etc...
And it is a fight with the borrow checker because there a plenty valid and correct code fragments that the borrow checker rejects.
And sometimes your friends are wrong because you know things they want, so you do what you were doing anyway.
I've not yet had the pleasure of using Rust, but you've described my experience of recapitulating ML-style functional programming in terms of the C# type system.
But in reality it is overly conservative and you have to expend effort just to get things to work.
When that happens it feels a lot more like fighting a war, a waste of resources that ideally would have been avoided.
Of course this may improve in the future, but right now I would leave the rose-tinted glasses in the drawer.
See also threads like this: https://internals.rust-lang.org/t/accepting-nested-method-ca...
{
let mut x = 1;
let y = &x;
do_stuff_with(y);
x += 1 // not allowed right now because y is still active
// no use of y here. possible uses of x
}
As it was, I felt as though I were manipulating a child in order to get it to do what I wanted.
But yeah, we should improve this.
Just pointing out that this isn't as major a problem when actually using Rust once you've picked it up.
I found once I grasped the model that the borrow checker uses that I rarely fight it and instead find it catching many things that I miss.
Right now, the rules are very simple to explain. But that means your code gets a little more complex at times.
And more awkward in places due to convincing the checker that what you're doing is ok.
It's initially quite intuitive and makes sense. But as you use the language more you start questioning why it doesn't make sense and you eventually have to learn how rvalues work.
Atleast steveklabnik had the good sense to acknowledge the issue rather than blame it on the developer.
I'm merely pointing out that this is not always an issue when programming in Rust, just mostly for newcomers, which means that if you're willing to bear with it for a while whilst learning it won't be a problem later. This isn't an ideal situation and is one we should fix. But it's not as bad as a problem that plagues you whenever you are writing Rust code.
Some of these problems persist as awkwardness in even idiomatic code. Most go away. I write Rust code every day, and I've seen or written borrowck-induced awkward Rust code maybe twice this month.
All of this is not to say that it isn't a problem we should be fixing -- it is and we are. But it's less of an annoyance in everyday Rust programming than one initially gets the impression it will be.
Which is exactly what I was describing, why are you continuing to argue against the point?
I'm not implying that the problem is with the developer. I strongly feel that this is something we should fix. I strongly feel that programmers -- new or otherwise -- shouldn't have to deal with this.
I've said all this multiple times, go back and read my comments.. You're the one arguing the same point over and over.
I'm just pointing out that with some experience it ceases to be a problem. My point is simply this: This is not a problem that will plague you once you've been programming Rust for a while, so it's not one of those persistent annoyances that will stick with you no matter how much you program in a language. I'm only making that distinction. That it's not that bad. It's bad. We're fixing it. It's not the developer's fault. But neither is it a problem that will always plague you. It goes away as you start writing more idiomatic code. That should not be necessary, agreed, and we should fix that, but it is true.
I said as much in https://news.ycombinator.com/item?id=13415585
I'm starting to feel you're not commenting in good faith here. I'm not going to repeat the same thing over and over.
You claim it's not the developers issue all the while claiming the problem goes away once you get more experienced with Rust (hint: that means it's a developer issue).
What you really mean is that this issue affects people's non-local code design and that people will eventually get enough experience to head off the issues before they crop up.
What I'm saying, and what others have said, is that's a problem with rust and it needs to be fixed so it stops affecting such design.
I have found myself rewriting entire functions to get rid of awkwardness due to this issue, and having the experience to write it that way the first time doesn't mean it isn't an issue, and it doesn't mean that it should be idiomatic rust.
You keep trying to push this onto the developer, even while claiming you're not doing that.
This is _not_ shifting the blame to the developer. This is accepting the blame in the right place. The blame has been accepted. The rest of the point is not about blame (which has already been attributed), it's about hope.
Having accepted the blame, I'm saying that one should not feel too discouraged by this issue because most people eventually pick up a programming style that's less prone to this.
I'm not saying developers should have to. I'm not saying it's their fault that they haven't. It's totally our fault, and we should fix it. I'm just pointing out what usually happens for most folks learning Rust, so one should be optimistic about this problem and hopefully not be too discouraged by it :)
You're interpreting "The developer can avoid this by improving X" as a way of shifting blame to the developer. That is not the intention. It is merely pointing out that the developer can avoid the problem; the ability to avoid the problem does not imply that they are to be blamed for it. It's a comment to instil optimism about the issue. The blame for the issue has already been attributed.
I don't disagree that this is a problem with Rust. I don't disagree that Rust should fix it. It's being worked on. In the meantime, I'm advocating optimism for newcomers hitting this issue. One can do both.
Regardless of intention, it's a reasonable interpretation. Your initial response came across as defensive.
This particular issue was the final straw that made me decide to stop using Rust and wait. It makes the language feel sophomoric and not ready. I'm glad you guys are fixing the issue and maybe at some point I'll try Rust again but probably not before the borrow checker gets more sophisticated.
The ones who are not emotionally involved in Rust are not going to simply be hopeful that they'll eventually learn how to work around the deficiencies in the borrow checker.
Either way, we're beating a dead horse at this point so we may as well drop it.
Hope you find the time/inclination to pick up rust again! I'm sure these issues will be fixed eventually (hoping that is soon)
Others are due to not-immediately-obvious issues with code. One of the most common borrowck complaints I've seen is about the fact that you can't have mutable and immutable aliases operating simultaneously, even if you don't have threads involved. There's a reason behind that (http://manishearth.github.io/blog/2015/05/17/the-problem-wit...).
Take http://manishearth.github.io/blog/2015/05/17/the-problem-wit... for example...
fn main() {
let mut x = 0;
let y = &x;
println!("{}", y);
x += 1;
println!("{}", x);
}
What I do mean is, many times we have people drop by IRC and say "the borrow checker is dumb, I want to do X, and it won't let me" and the reply is something like "have you considered threading" and they go "ohhhhh I see." There was a good quote of the week about this at some point, but I can't find it right now.
I don't happen to run into patterns like the above code in real Rust code, personally. That doesn't mean that they don't happen, but I'm interested in _real world_ cases where this becomes an issue. We've got a bunch of them, which is why we're pursuing non-lexical lifetimes, but it seems like for many, this is a _constant_ struggle, and I run into a NLL situation maybe once every two months. Figuring out that disconnect is very interesting to me. It matters because I'm going to be responsible for teaching it, so real examples are way better than contrived ones. Moving that focus is the point of my current re-write of these docs :)
Your argument that for real world programs the borrow checkers conservatism does not create technical problems is completely plausible to me. The issue lies somewhere else.
There is a psychological component to this. People struggle with the borrow checker, or at least they feel like they did. This is negative for rust.
My point is that it would be better if it weren't trivial to write short, safe programs that are not accepted by the compiler.
Where can one expect the line to be drawn? I don't know. I worked with Isabelle/HOL to formalize mathematical proofs and there I saw that we can indeed expect a computer to be reasonably clever in abstract reasoning. I'm not saying that the rust compiler is dumb, however, there is room for improvement.
Thanks for your(plural) work on rust, btw.
(I even tried to get unsafe renamed to "trusted"... that didn't work out.)
Recently I was parsing a line-oriented, whitespace delimited file and had to introduce a temporary ("line", below):
let f = File::open(path)?;
let reader = BufReader::new(f);
for line_result in reader.lines() {
let line = line_result?;
let mut word_iter = line.split_whitespace();
For me, it seems like every time I dip a toe into rust I run into borrow/NLL issues right off the bat.
Ah, see, I'd go straight for Entry in the hash case. Which might be what you're referring to.
yeah, that later one isn't NLL: it's that
line_result?.split_whitespace()
^ ^
Thanks for the clarification about the second case not being solved by NLL. Those examples were the ones I had in my mental "code I would like the borrow checker to accept" bucket.
Would Niko's proposals in that internals thread you linked ("Accepting nested method calls with an `&mut self` receiver") eliminate the need for the temporary in that second example?
let s = String::from("hello");
let s = &s[..];
let s = String::from("hello");
{
let s = &s[..3];
// here, s is "hell"
}
// here, it's the full string again
Why do I need to manually create a scope when the compiler can see that variable is never used again afterwards?
In Jill.name().first_name(). Why can't the compiler extend the lifetime of the return value until the end of the statement?
Having a borrow check to ensure memory safety is good. But those are just unnecessary and pointless. These are not pitfalls in user code, but pitfalls in language design
If you look at the rest of this thread, you'll see that the compiler _can't_ currently see that; this is the whole issue.
1. For a given school, return all the students
2. For a given school, return all the classes
In the last data model given (the correct one), the above 2 things can't be done (I think) without doing something funky like enumerating all the Enrollments and finding unique students/classes. In order to modify the data model to provide the needed information, wouldn't you run into the same borrow checker problems?
Edit: To clarify, I am a Rust neophyte and have only begun to scratch the surface of the language.
Edit2: Also, "There is no reason a person should know which class they are in, nor should a class know which people are enrolled in it. "
I know what the author is trying to say but when you read that at face value it comes across as a little humorous.
One problem is that the user would have no way of naming the thing that the lifetime is attached to. Diagnostic messages when trying to hand off 'name' to other code could be perplexing, as the delta between broken code and working code would be larger.
28 | let name = jill.name().first_name();
| ----------- value returned here
28 | let name = jill.name().first_name();
| ^^^^^^^^^^^^^ leaves scope before use,
| and is unavailable hereIt could, but it doesn't yet. This kind of analysis needs porting borrowck to MIR, which still has to happen. It's too complicated given the current way it works.
Though it is a matter of explicit over implicit too. If `name()` had a destructor you may want it to be explicitly tied to a scope by dint of being declared in one instead of floating around in a magical hidden variable that's tied to the scope of a different one. So there are reasons for why it may never be made to work that way.
I'm not sure I see what the problem is with this. It seems like it is very intuitive behaviour in this case and exactly what you would expect.
This is similar to vvanders' and steve's comments in https://news.ycombinator.com/item?id=13413084 ; such things lead to a more complicated mental model even if on the surface it looks simpler.
Also replacing a O(1) chasing a pointer into a O(n) array search