Still in love with Rust

by adamnemecek • 8 years ago

183 181 8 years ago

I'd still love to see a good tutorial on setting up Vim to work well with rls and all the other plugins. I only really keep VSCode around since it works with rls and sometimes I want to use the autocomplete for codebases that are large enough to really need good autocomplete, but not so large that rls gets bogged down and doesn't work at all anymore.

vbezhenar• 8 years ago

Idea worked wonderfully for me with tiny pet projects.

lkurusa• 8 years ago

Does vim-racer[0] not work for you? I don't exactly know what you mean by "good autocomplete", but if you are more specific, then maybe we can help you a little more? :-)

FWIW, my Rust setup in vim is mostly the following plugins:

* NERDCommenter, for commenting stuff

* rust.vim, for general rust things (like integration with rustfmt, playground, and the next two plugins)

* tagbar, for showing the list of tags (of the ctags-esque sense) in a sidebar.

* syntastic, for on-the-go syntax checking.

[0] https://github.com/racer-rust/vim-racer

bhengaij• 8 years ago

Same. Rust needs a one-click install "integrated de"

mcjiggerlog• 8 years ago

For language server autocompletion, install coc.nvim - https://github.com/neoclide/coc.nvim

Then

  :CocInstall coc-rls

and should work out of the box.

I also use https://github.com/w0rp/ale to provide linting. Also works out of the box.

You can get these things running on vim8, but honestly it's easier to just use neovim.

If it's of any use, you can check out my full config at https://github.com/tom-james-watson/dotfiles.

onhn• 8 years ago

“Bad programmers worry about the code. Good programmers worry about data structures and their relationships.”

I think this hits the nail on the head for me: abstract structure first, the choice of language C/C++/Rust/... is then a secondary consideration.

pepper_sauce• 8 years ago

I don't think that holds, given the potential scale and severity of memory safety bugs in systems programming.

I'm not saying Rust is the right answer but not paying attention to, or worse, putting our heads in the sand, when it comes to our tools is not the answer.

_i_think_• 8 years ago

let me tell you that Rust forces you to very carefully consider data structures and their relationships.

The point of a programming tool is to help the programmer to be a better programmer, there is indeed no point using a very strict language, statically typed with lot of constraints attached for a perfect programmer.

But for commoners it would certainly help to have some guiding.

a_imho• 8 years ago

Which brings up the question when/why to pick Rust? There are a plethora of languages which from my layman perspective seem more interesting. OTOH there are a plethora of languages with rich ecosystems, proven pedigree and with a high demand for such developers.

seren• 8 years ago

I am working on medical devices, where a bug can have catastrophic consequences.

Essentially the ecosystem today is C and C++ for embedded FW and middleware. Even if we are careful, we rely on static analysis tools to avoid some categories of bug. (Plus a humongous amount of tests). Typically bugs that would be caught (mostly) by the Rust compiler at compile time. I think it is a great idea to have some safety guarantee at compile time.

Would Rust fixes every problem ? Absolutely not, you'll always get logical bug or implementation not conform to the requirements but I think it is a move in the right direction, and I hope that slowly in next 10 years Rust will grow in the embedded space.

And overall cargo is nice, C/C++ don't have standard package manager. The test infrastructure is there by default, this is a breeze to add them in your code, rather than relying on an external framework, so this is also good.

I don't really see an alternative today for small footprint SW where C and C++ is king, and I hope that embedded toolchain vendors are taking notes.

If today you mostly code in javascript or python, unless you need bare metal like performance or to have stronger safety guarantee, maybe it does not make sense to move to Rust.

adrianN2• 8 years ago

Sure, nobody would choose Rust over C++ if you couldn't express pretty much the same data structures and relationships in Rust that you can express in C++.

hasahmed• 8 years ago

Rust sounds like the dream. I gotta check it out sometime

hbbio3• 8 years ago

Rust is a wonderful language, but I still have the impression that it is not stable as of 2018. All the toys I've made to play with it during the last years went deprecated quickly, especially if you rely on the ecosystem of packages (web server, database, ...).

In addition, there was this thread: https://internals.rust-lang.org/t/concerned-about-rust-2018-...

All in all, Rust will be great when stable (including the ecosystem).

microtonal3• 8 years ago

I think it depends a bit on which in part of the ecosystem you work. I do not do web applications at all, but use Rust for machine learning. Things like ndarray, petgraph, and the Tensorflow bindings have been very stable for me.

The only large change that I had to make over the last year due to ecosystem changes was going from error-chain to failure. Of course, this was not strictly necessary, but failure seems to be more popular now and I generally prefer it over error-chain.

rqs• 8 years ago

depends a bit on which in part of the ecosystem you work

This is also my experience. The feeling towards Rust is highly depends on the dependency of choice.

Rust itself is mostly very stable though. I've implemented few pure algorithms in Rust a little over two years ago, and it still compile and runs today without having to change a single letter.

Wish one day I can say the same thing to my other Rust projects :)

the_mitsuhiko• 8 years ago

All the toys I've made to play with it during the last years went deprecated quickly

In which sense? That dependencies evolved?

hbbio3• 8 years ago

Languages features, dependencies, etc.

For instance one project:

   error[E0010]: allocations are not allowed in constants                          
   error[E0015]: calls in constants are limited to tuple structs and tuple variants

Another:

    error: could not find native static library `brotli`, perhaps an -L flag is missing?
                                                                                
    error: aborting due to previous error                                           
                                                                                
    error: Could not compile `brotli-sys`.

And I'm not talking what happens if you don't fix versions in Cargo.toml.

simias• 8 years ago

What do you mean by deprecated? Unsupported dependencies? All the code I've written post-1.0 still compiles correctly (although a lot of it could use an update to use new language capabilities).

I even have a bit of Rust in production at work and I haven't encountered any maintenance issue so far.

Now the ecosystem itself can move pretty fast depending on the dependencies you use, that's true, but that's a different issue. You're never forced to update if you don't want to, you code keeps compiling with newer Rust releases. Compared to most languages I've experienced with, Rust has been very stable post-1.0. And they seem to take that issue very seriously.

The thread you linked might have some truth in it but it seems like the main point is "look, you've had a .2 patch release!" That's not a good thing of course but unless you happened to jump on that new functionality as soon as it got released the impact is rather minor and bugs happen.

sheepmullet2• 8 years ago

Now the ecosystem itself can move pretty fast depending on the dependencies you use, that's true, but that's a different issue.

Is it? I’m used to languages which bundle a huge set of standard libraries as part of the language.

Svip2• 8 years ago

Thank you for this piece. I am working on a game, that I am prototyping in Go. In my prototype, I have just entered the realm of multiple threads, and for this project to work out, it needs to handle quite a lot of threads working on the same resources.

While I have not personally dappled this far into concurrency in Go myself, I have read plenty of articles where developers recommend using the mutex package rather than the goroutines themselves.

But I've also read about Rust's thread and resource handling -- and I'll admit, I've been looking for an excuse to write some Rust again -- and so inspired by this entry, I've decided to finish the prototype in Go, and build the real version in Rust.

monocasa• 8 years ago

There was a great talk recently about how to properly build an entity component framework in Rust if you haven't seen it yet.

https://youtu.be/aKLntZcp27M

Svip2• 8 years ago

Thank you for that link. Very useful. It already highlighted a solution in Rust for a problem that I've run into with my Go prototype. And I appreciate there also was a blog post to read.[0] I personally prefer articles over videos.

[0] https://kyren.github.io/2018/09/14/rustconf-talk.html

vaylian2• 8 years ago

Rust will force you to be a good programmer,
[if] you like it or not.

This is probably the best in-a-nutshell statement that describes what a good programming language is for me.

I had similar moments in the past. Before Python I cared about indention to some degree. But once I got used to the way Python forces you to indent your code, I came to the realization that this is pretty much the way I should format my code anyway and it influenced my coding style in all other block-based languages for the better.

Clojure was another language that showed me how functional programming should be done. My code in all languages changed and I used a lot less variables and a lot more constants and parameters. My code and the systems I created became a lot simpler and much more robust as a consequence. I had used Haskell before but it never really transformed my mind the way Clojure did. I also got a really good grasp on destructuring and list comprehension.

Rust offered me several relevations. I finally understood test-driven-development because the language has excellent support for it. I also understood RAII more deeply. Traits became clear to me (Clojure protocols and Haskell type classes felt strange to me). I finally understood why the Maybe monad in Haskell/ML is one of the simplest and yet bests ideas in programming.

And there is the borrow checker. I already had a functional programming mindset when I picked up Rust. But having a clear owner of data takes the mindset of functional programming even further[1], even when mutability is involved. It is a lot easier to reason about your code when you know exactly which operations can (and should) change a variable.

An aspect that the Python, Clojure and Rust communities have in common is that they emphasize idiomatic code a lot. Therefore it is much easier for newcomers to find examples of professional-level code and learn things the right way. And it is much easier to dive into a new codebase. Clojure is actually a very extreme case of this, because Clojure code tends to use a lot of 1-letter parameter names. In most programming languages you are allowed to have the variable "i" for looping over numbers. But not much more. In contrast, Clojure has a list of well-established 1-letter variables that are recognized by everyone in the community.

[1] This does not adhere to the classical definition of functional programming but I feel that it is a natural extension to it.

goldfeld4• 8 years ago

Python formatting is egregious though, with hopelessly long lines and no sign of where to break them up, or at the most weird places. Yes I'm one of those persons who adheres to the terminal 80 column rule, it's neat for splitting and cascades into many benefits. Python often looks like unkempt code to me, like the developer has no care for how it looks or layouts (which isn't true, it's Python's fault.) The forced indentation (I love proper formatting with consistency and indentation in languages) is among the first reasons I never liked or used Python, until recently when I gave in because it's such a great ecosystem to build scripts and such, but I still hate the formatting everyday. Clojure is beautiful though. Even C is gorgeous compared to a Python 'forced' program layout. I'll start adding Hy to the mix and touch vanilla Python as little as possible, but I still have to read library code.

growtofill• 8 years ago

Doesn't Python have an autoformatting tool like gofmt or Prettier?

vaylian2• 8 years ago

I follow the 80 columns rule as well. Do you have an example of such unkempt code that you can share?

freyir• 8 years ago

Python formatting is egregious though, with hopelessly long lines

This is the result of writing hopelessly long lines of code. Python doesn't force you to write long lines.

I've done it myself, many times, but gradually made efforts to avoid this. I'll move deeply indented code into a new function or split up a long expression with a temporary variable. If I run into a particularly hard-to-format portion of code, it's usually a sign of that my code could be better.

int_19h6• 8 years ago

Long lines come most often in conditions of if- and while-blocks, mostly because there's no way to split them over multiple lines that isn't visually hideous.

But a condition can be easily split if it is assigned to a variable, and that variable then tested. And naming said variable well can make a comment explaining the condition redundant.

sifoobar5• 8 years ago

Only as good as Rust allows you to be, which is fine for some until they stagnate or grow out of it; but far from the final answer to anything.

I don't get at all how being forced to do anything could ever be a good thing. Smells like cognitive dissonance from here. I'm all for powerful tools that enables me to write better code faster; but being forced, really? That's the best thing about Rust? Ew.

knocte• 8 years ago

I don't get at all how being forced to do anything could ever be a good thing

If you don't prefer tools that can automatically check if you were a good developer or not, it not only does not scale (to trust the software you build), it's that I wouldn't want to work with you as a teammate.

majewsky4• 8 years ago

I don't get at all how being forced to do anything could ever be a good thing.

So anarchy it is, then?

Jach• 8 years ago

I'd chalk it up to the old advice of "learn new programming languages that teach you something new [because you'll become a better programmer]" applying to Rust, as there is something new to most programmers in it. I remain skeptical that a total beginner starting with just Rust will be any better of a programmer (however you're measuring that) than another beginner who had started with, say, Python, giving both a year.

recursivelambda• 8 years ago

So all programming languages at a higher level than ASM are horrible to you? They don't allow you to write everything that pure ASM does. Even C enforces structured programming.

unrealhoang• 8 years ago

Have you ever heard about the new thing called static typing programming language? I heard it force people to put only data with correct type into a function, it doesn't allow me to put any memory address I want to, such bad thing. Ew.

Jokes aside, I see rust borrow checker is just another dimension of type checking, it is type checking for access behavior and that's all about it.

Just as you can wrap your data in memory into static type system and leverage it to make a class of mistake impossible (passing wrong data to a function), you can wrap your access behavior into lifetime type system (converting raw pointer access behavior to reference with correct lifetime/access using unsafe) and make a class of mistake impossible.

In my definition, that's a powerful tools that help me to write better code faster.

ElBarto6• 8 years ago

A programming language won't make you a good programmer in the same way as a brush won't make you a good painter.

steveklabnik21• 8 years ago

I'm not aware of any paintbrushes that refuse to apply any paint unless you hold them in a certain way. They may apply it poorly, but you're still getting some paint on that canvas.

I think a more proper tooling analogy is safety tools, like the Saw Stop. They don't guarantee that you make a good construction, but they do either prevent or significantly reduce the chance of you losing a thumb.

Sharlin2• 8 years ago

A language (programming or otherwise) is much more than a just a simple tool. It's a way of thinking. A language that guides you toward good solutions definitely makes you a better programmer if you give it a chance.

EugeneOZ11• 8 years ago

"No race conditions, leaking resources, dangling pointers, unhandled exceptions, ..., the list goes on."

Except of dangling pointers, Rust has everything else from this list.

About compilation time issue: it doesn't exist anymore. Current versions do "cargo check" in a few seconds even for big projects (my biggest project is 45k LoC) and it takes a couple of minutes to compile it because of incremental compilation. For libraries it takes seconds to compile and 1-2 seconds to run "cargo check" (or clippy).

Laaas• 8 years ago

It is indeed true that you can unintentionally leak resources, but it is still much harder. Also, what do you mean that it has unhandled exceptions and race conditions? Rust doesn't even have exceptions (although there are panics, which is not the same thing), and race conditions are statically prevented in safe code due to how a mutable reference works.

EugeneOZ11• 8 years ago

Theoretically they are different, in practice they are the same (it's just more difficult to handle panicking - sometimes you can't use catch_unwind because of it's trait limitations).

About race conditions: https://doc.rust-lang.org/nomicon/races.html

Sharlin2• 8 years ago

It's important to understand the difference between data races and race conditions. The former are a subset of the latter. Race conditions in general can happen even in single-threaded code. (Although admittedly Rust ownership semantics prevent many of the latter as well; for instance, modifying a collection while iterating over it is impossible, cf. Java `ConcurrentModificationException` which itself is thrown only on a best-effort basis.)

jerrre4• 8 years ago

big projects (my biggest project is 45k LoC) and it takes a couple of minutes to compile it

I don't think 45k LoC could be considered big, and I _do_ think a couple minutes compile time is a problem. I say that as a C++ programmer, a full recompile will easily get me out of the zone

EugeneOZ11• 8 years ago

Out of curiosity, not to argue: how much time it takes in C++ to compile 45k LoC?

pjmlp4• 8 years ago

It surely does, my Gtkmm demo application I wrote several years for "The C/C++ Users Journal" still compiles faster than the Rust rewrite in Gtk-rs, when doing fresh build or after minor changes.

Lack of binary library support on cargo, not having incremental compilation and linking does hurt.

EugeneOZ11• 8 years ago

There is incremental compilation. And cargo doesn't recompile libraries every time. Not sure how you don't know it if you really use Rust.

z3t42• 8 years ago

Can't have a Rust post without a mention of D.

geo_mer• 8 years ago

Native async/await support is really great. I haven't gotten very far with Rust but it already made me rethink many things that I thought I had understood.

LukaD• 8 years ago

The comment about "no race conditions" is not accurate. Rust guarantees an absence of data races which are not the same as race conditions.

https://doc.rust-lang.org/nomicon/races.html

kungtotte2• 8 years ago

Another point against Rust is binary size.

If you write the Fibonacci generator in one of the first chapters of the Rust book it will create a 4MB binary. Even stripped it's over 400KB.

If you write the same code in Nim it compiles down to 112KB with debug symbols, unstripped. Statically linked against libc (still with debug symbols) it's only 850KB.

850KB statically linked versus ~450KB dynamically linked and stripped.

For code that is virtually identical.

microtonal3• 8 years ago

Rust binaries include jemalloc. However, jemalloc is removed in nightly and by default Rust will use the system allocator:

https://internals.rust-lang.org/t/jemalloc-was-just-removed-...

This should reduce binary sizes quite a bit.

vesak3• 8 years ago

And some rough numbers, rustc-1.30.1 vs rustc-1.32.0-2018-11-21, building a hello world with --release...

Mac OS:

- 1.30.1 before strip: 573K, after strip: 380K

- 1.32.0 before strip: 267K, after strip: 178K

Ubuntu 18.04 LTS:

- 1.30.1 before strip: 3.9M, after strip: 407K

- 1.32.0 before strip: 2.3M, after strip: 191K

weavie• 8 years ago

How far does this scale? What are the binary sizes of a 100k loc app in Rust vs Nim?

nimmer3• 8 years ago

See reply from kungtotte

adaszko• 8 years ago

Does this matter anywhere outside of embedded software nowadays?

eeZah7Ux2• 8 years ago

Yes: security sensitive applications where being able to inspect or fuzz a binary is strongly dependent on size. Also firmware on servers.

jerrre4• 8 years ago

Let's face it – typically developers are familiar with OOP, garbage collected, dynamic programming languages.

I think most of these programmers, and their project are not the target of Rust? If you're writing Java/C# or higher level languages, you already made the choice to sacrifice some computing efficiency for programming efficiency.

I always thought Rust was more destined to convert C and C++ (and D/Swift/Go ?) programmers, that need the raw power and control.

projektir• 8 years ago

I think most of these programmers, and their project are not the target of Rust? If you're writing Java/C# or higher level languages, you already made the choice to sacrifice some computing efficiency for programming efficiency.

Or you just don't know any better, as was the case for me.

There are many benefits to Rust that are not related to runtime efficiency.

otikik2• 8 years ago

I don't know the demographics/targets, but I come to rust from ruby. So does Steve Kabnick, the author of the Rust book.

jerrre4• 8 years ago

Cool!

What kind of projects do you work on, and what made you switch?

steveklabnik21• 8 years ago

We have three major audiences:

* "systems programmers", aka the C and C++ folk

* "functional programmers", largely Haskell folk

* "scripting programmers", mostly Ruby/Python/JavaScript folk

Each has gotten something out of Rust, and also brought some things to Rust.

int_19h6• 8 years ago

You're missing the group which Java, C#, Swift, D and Go co-inhabit. Somewhere between "systems" and "scripting".

Tade0• 8 years ago

I originally picked up Rust because I couldn't make audio processing in JS fast enough.

I stayed because I found Rust to be a language that not only doesn't get in your way, but also helps you out of some situations.

I mean, I had the compiler say to me "Perhaps you could [...]". Perhaps. And nine out ten times it was a good suggestion.

leni536• 8 years ago

I tried Rust a few weeks ago. Is my understanding correct that Rust doesn't have the equivalents to the following C++ features yet?

- non-type and template template parameters

- constexpr functions

I understand that Rust macros can replace them some of the time, and they are much better then the C preprocessor, but that's quite a low bar to meet. Last time I checked non-type template parameters (called "const generics" in Rust) is on the roadmap, so it is promising.

th0br0• 8 years ago

A `const fn` RFC is accepted and somwhat available when using nightly compilers, also see https://github.com/rust-lang/rust/issues/24111

steveklabnik21• 8 years ago

The equivalent of "constexpr functions" has a design, and is usable in nightly.

The equivalent of "non-type template parameters" has a design, but is not yet implemented. It's being worked on.

There's no direct design for the equivalent of "template template parameters", but a similar feature has a design, but is not yet implemented. It's being worked on.

arendtio3• 8 years ago

Barriers to entry

So far I had two faltering attempts to learn Rust but somehow I didn't find good enough material to get me hooked.

I am looking for something similar to the Tour of Go[1], where you can interactively try the language by solving minimal tasks. Does someone know of such material?

[1] https://tour.golang.org

bugmen0t• 8 years ago

https://github.com/carols10cents/rustlings

steveklabnik21• 8 years ago

https://doc.rust-lang.org/rust-by-example/

arendtio3• 8 years ago

That looks pretty good. Thanks :-)

majewsky4• 8 years ago

I found The Book [1] to be pretty good, but it suffers the same problem that I find infuriating about programming language books in general. They have to target one specific proficiency level, so someone less experienced is going to get left behind, and someone more experienced is going to get bored out of their mind.

As a somewhat experienced Rustacean, my biggest frustration as of now is the lack of a comprehensive reference where I can look up very specific things (e.g. the syntax for byte string literals, or an exact definition of how borrowing works in some particular edge case). There is [2] but as it says on the first page it's incomplete.

[1] https://doc.rust-lang.org/book/2018-edition/index.html

[2] https://doc.rust-lang.org/reference/introduction.html

kieckerjan• 8 years ago

I come from C. What got me started was the O'Reilly book, Programming Rust by Blandy & Orendorff. (It cultivates the your sense of snobbery too. Imagine code examples using lists of "murderous Renaissance artists".)

arendtio3• 8 years ago

In fact, that was the book I started with once. I kept reading until page 76. The book isn't bad, but I think books are just not the best medium to learn new languages, at least for me.

Vogtinator• 8 years ago

Am I the only one who really can't stand rust?

I was and still am very much in favor of the ideas and concepts, but when I actually tried to use it my opinion on it turned by 180°:

- The syntax is awful - No dynamic libraries - Slow compile times - Many checks are too stupid to figure out some valid cases (maybe that improved meanwhile)

steveklabnik21• 8 years ago

You are not.

- No dynamic libraries

This is technically incorrect, but practically correct, by the way.

(maybe that improved meanwhile)

The next version, released on December 6th, ships "non-lexical lifetimes", which does make the borrow checker smarter.

_pmf_• 8 years ago

No. I wanted to like it, but a lot of stuff (standard type names, syntax) is different just for the sake of being different.

I wish D would have won the C++-successor race.

vesak3• 8 years ago

No dynamic libraries

While I understand that some people prefer this, I prefer static linking to dynamic linking. So much in fact that I refuse to use environments that only do dynamic linking in any of my projects.

Yes, I have to make a few concessions to that because of the idiocy of glibc, but other than that.

andrenth• 8 years ago

I have really tried to like it. Coming from OCaml, it has many of the things I’ve come to expect in a modern language, but with better tooling and more libraries.

I generally do not mind syntax. I like OCaml’s and Erlang’s, for example, which are usually considered ugly, but for some reason I can’t enjoy the look of Rust code. I think all the nested <<<>>> and the use of :: are a large part of this.

But the real issue for me was refactoring owned struct fields into borrowed ones. The ‘a lifetimes leak everywhere; you have to add them to the struct field, after its “impls”, and everywhere the struct name appears.

I’ve found that Pony provides the same level of safety in a much more lightweight manner with its capabilities system. Interestingly, if I’m not mistaken, early Rust was somewhat similar to Pony.

Unfortunately Pony doesn’t have the same development resources and thus the ecosystem is somewhat lacking.

jjgreen• 8 years ago

I was going to try out Rust, then I saw this on the download page

  curl https://sh.rustup.rs -sSf | sh

Really?

steveklabnik21• 8 years ago

Yes.

The script is less than 400 lines, and is pretty easy to audit. We wrote it in such a way to prevent some issues like "if the download gets cut off only part of the script gets run." It's served over HTTPS.

You can also probably get an install from whatever package manager you use, if you prefer that.

tpxl• 8 years ago

You can detect whether the script is being piped into sh, so downloading the script via for example FF might produce different results than piping it to shell, making audits useless.

ElBarto6• 8 years ago

I've been looking at Rust recently, and my conclusion is that it is not production-ready.

It seems like an ever-changing work in progress at the moment, with idiosyncrasies such as this.

woile2• 8 years ago

How does data-centric work? How is it different from OOP? Can someone elaborate for a person using python?

Also, what opinion rust devs have over golang?

steveklabnik21• 8 years ago

How does data-centric work? How is it different from OOP?

Rust doens't have objects. You have structs, which are purely data, and functions, which are purely behavior. OOP puts the two together. So in Rust, you don't tend to design things by saying "what are the objects I need?" but rather "what is the data I am manipulating and how do I manipulate it?"

Also, what opinion rust devs have over golang?

I am not 100% sure what you're asking here, could you maybe re-phrase? Are you asking why Rust is better than Go?

woile2• 8 years ago

Oh nice, good to know.

Are you asking why Rust is better than Go?

No, just the opinion in general, why rust over go? I know go is more oriented to distributed systems, but it seems like rust eventually will be used for that as well. And I also see golang being used to build just anything, like gopass for example.

snrji• 8 years ago

Rust still lacks an effect system, though. Any plans?

steveklabnik21• 8 years ago

No current plans.

axilmar• 8 years ago

The moment a programmer conflates OOP with 'complicated graphs with lots of interdependencies between nodes', it's the moment I stop reading their writings.

OOP does not force one write complicated graphs of objects. OOP has its applicability, FP has its applicability and well-written code in conventional languages is a mixture of those two.

insulanian2• 8 years ago

Can someone familiar with both Rust and F# (or OCaml) type systems write few words about the expressiveness aspect of the language?

What would I miss in Rust if coming from the F# background?

steveklabnik21• 8 years ago

Fun fact: Rust was originally implemented in OCaml, and takes a lot of inspiration from it. Rust was originally described to me by someone as "C meets OCaml."

The biggest thing missing in Rust that are in both are higher kinded types. The biggest thing missing from OCaml is parameterized modules.

insulanian2• 8 years ago

F# doesn't actually have HKTs either, unfortunately.

mamcx• 8 years ago

In the surface you can say yes or no (only looking at types and such). But summing all the features, F# is far more immediately productive.

I take rust after 2 years on F# (that I continue to use). I have like 2 months of Rust now. I still struggle to solve things that are "no-brainers" in F# or swift (mainly, because Rust is VERY strict about know statically all about the code, and VERY hard to build generic, dynamic code).

So, it will look less expresive, IMHO. I still ramping-up and start to get more a more of the rust style of programming, so the pain is reduce. With the additions for rust in the near future it will reduce it more, I hope.

millstone• 8 years ago

Now I see how before Rust I was often writing my code in a typical OOP-way: forming complicated graphs, with many objects referencing each other, with unclear relationships. The class APIs, inheritance hierarchies, were there, but they were just obstructing the core problem: structuring and manipulating data... Rust will force you to be a good programmer, you like it or not.

Rust today demands static knowledge of all relationships. But class APIs, inheritance hierarchies, and the like are powerful tools for dealing with dynamic relationships: code that runs together yet evolves separately, where Rust is feeble. Good programmers recognize that different problems require different solutions!

Dowwie3• 8 years ago

Rust is feeble at patterns it wasn't made for (OOP), absolutely. :). Composition over inheritance.

Dowwie3• 8 years ago

dpc, the author, is such a helpful, positive force at the #rust-beginners IRC channel

Dowwie3• 8 years ago

Rust core development is working on many important features in parallel. The feature I have been most excited about, and wish was here already, is async/await. I wrote quite a bit of asyncio using futures and it was a bloody, exhausting experience. Lack of documentation or examples made the experience even more challenging. Much of what I wrote could be trivially refactored to async/await syntax, but I may keep some around to honor hard-earned gains. It's not clear to me yet whether the first stable release in Q1 2019 will allow me to port everything but I look forward to finding out.

purplezooey• 8 years ago

Bad programmers worry about the code. Good programmers worry about data structures and their relationships.

Mildly horse shit. It takes both on a team.

Still in love with Rust

Related Stories