Hacker News new | ask | show | jobs
by Kranar 1091 days ago
Four months to become as proficient in a memory safe, data race safe, and high performance language as in other languages seems like an astonishing accomplishment.

In fact it's frankly unbelievable, I'd have to imagine these guys are coming from a C++ background.

2 comments

Anecdotally, C++ developers often have a harder time coming to Rust than folks from a more scripty background. They have to unlearn things, and that can be harder than learning things.

You can see similar opinions expressed in this thread, like here, for example: https://news.ycombinator.com/item?id=36496654

Python programmers love Rust: - Lots of compilation errors, but then the code does what it says - Fast! That gives an immediate benefit that people coming from C++ don't see.
Really? Coming from C++, learning Rust felt very natural and welcome because it put a name and a structure to the mental model I had been using to keep a semblance of sanity while doing C++. Maybe learning primarily with C++11 was a factor?
It's a broad strokes kind of thing, not a universal thing. To be honest I think it has more to do with attitudes towards the language and compiler that tend to be more prevalent in C and C++ circles.

That is, a lot of people expect that the compiler is a tool that must do what they say, not a collaborative partner. Back in the IRC days, people used to join, and be like "Rust won't let me do X. X is obviously okay. How do I get the compiler to shut up and do X?" and the reply would be something like "what if there was another thread? that's what Rust is saving you from" or "Rust's pointer rules are different than the language you're used to."

Whereas folks from scripty backgrounds don't have preconceived notions of "I should be able to do this in this way," and so tend to trust the compiler more. Heck, that's why a lot of them are learning Rust instead of C; they know the compiler can help them out, and C's compilers cannot to the same degree.

Now, that doesn't mean that scripty people don't struggle, or that C and C++ developers don't have their own advantages in learning. Just that I don't think it's straightforward to say who has the overall easier time. It's more of a "having learned something like C or C++ does not universally advantage you."

Ha. Given that the C and C++ compilers optimize your code by inferring undefined behavior and other crazy stuff like that, it's funny to imagine you're not just playing with just another code generator. :p

(In fact, if you decide to be responsible about what you're telling your poor compiler to do and use a `size_t i` instead of an `int i` in your for loop, since iteration can never go negative and, if overflow happens, overflow on a signed type is worse since that's UB, I think you just made your for loop slower since all the optimizations the compiler was going to generate from inferring that iteration won't be infinite because that would overflow the `int i` and that would be UB so it can just ignore the possibility that the iteration would be infite went out the window and...man I really need to switch to Rust.)

((I've pretty painlessly learned and love Rust, btw. I'm still using C at home lately though cuz... masochism.))

I agree with you, but many trot out the "portable assembly" line.
Thanks for the elaboration. Now I'm wondering if this hasn't something to do with bottom-up people vs top-down people and their natural affinity towards lower level or higher level languages.

I'm very top-down, always starting by sketching the API I'd like to see for the feature, and then filling in the implementation. A lot of people I know from C++ are bottom-up, they toy with the implementation and then go up to the API. I found that what they like in systems programming is being close to the machine, while my interest lies in the OS boundaries and interfaces, not really the hardware.

I'm thinking maybe those different ways of approaching programming make it easier or harder to learn rust.

I think that's possibly insightful, yeah. A similar effect: a lot of people say "Rust is bad for exploratory programming because the compiler gets in your way." For me, it's fantastic for exploratory programming specifically because when I change something, it gives me a list of the other things I need to change! That's huge! But for others, it seems to harsh their buzz. I don't know how to reconcile these opinions other than "they're opinions and different people are different."
I don't think it has to be this hard to have memory safety, data race safety, and performance. After building and doing PL design in this space for a decade, I don't believe the assumption that Rust's or C++'s difficulties are inherent.

I think C++ has its own legacy difficulties (which also make transitioning to memory safety tricky), and Rust's choice of borrow checking is only one (sometimes difficult) technique for getting these aspects. But there are almost a dozen other methods out there for getting memory safety besides RC, GC, or borrow checking.

I rather think that these other approaches aren't mature enough yet to enter the mainstream, and we haven't seen them yet.

How do you build a data race safe language without taking on the restriction of immutable data (which is imo, a much worse and bigger tradeoff than the borrow checker)?

I actually really like the borrow checker as a tradeoff, I think it makes code much easier to understand and it makes all aliasing bugs impossible. The removal of aliasing bugs is I think an undersold benefit of using rust.

There are a lot of ways, but I think the most promising ones involve regions. We do this a lot manually in Rust, but a language could make this a first class concept. Some examples:

* Vale combines generational references and linear types with regions to eliminate overhead.

* Verona lets you divide memory into regions which can be backed by either arena allocation or GC. I think this is promising because for most GC regions you can completely avoid the collections.

* Cone lets you put borrow checking on top of any aliasing memory strategy, so could be something like the best of all these worlds.

* No language is doing this yet, but RC plus regions to eliminate the refcounts, then adding in value types for better cache usage, could be a real winner.

On phone so it's hard to get links, but you get the idea. The nice thing about regions is that they allow composing borrowing and shared mutability, something that the borrow checker struggles a bit with. Regions let us alias freely, and then freeze an entire area of memory all at once. Not that Rust isnt a good approach (it is!) but there are some easier techniques on the horizon IMO.

Please demonstrate a practical and memory safe systems programming language without borrowing.

I'd be delighted to see it, because right now I am not aware of any practical way to have memory safe regions without static tracking of borrowing from these regions. It's either that or runtime checking.

This might be of interest: https://verdagon.dev/blog/first-regions-prototype

It uses region-based static analysis without borrow checking: it doesn't impose aliasability-xor-mutability per object, or even per-region.

Though, if you'd like to move the goalposts further to no form of borrowing at all, then I recommend looking at languages like Forty2 and Verona, they might be what you're looking for.

I have already spent more time than I wanted on reading through verbose but elusive articles about Vale, without any insight into how this actually happens.

I have already spent too much time trying to compile Vale compiler which is a weird mix of Scala and C++ with a small Vale driver. Once it is actually written in Vale without segfaults, I'll revisit the language again.

Thanks for the Verona recommendation.