Replies: 3 comments 15 replies
-
Rust has been fantastic for Deno. Zero regrets.
Absolutely. I'll never start a new C++ project. Rust services the same problem domain, but solves it better, in many ways. More than anything, having a single build system (cargo) and a system for linking in 3rd party code (crates.io) helps immensely. The biggest problem in C++ is linking to 3rd party code. Every project uses different systems: gn, gyp, cmake, autoconf, make. Rust is better in a myriad other ways too. But we really appreciate being able to pull in complex 3rd party native code without much trouble. |
Beta Was this translation helpful? Give feedback.
-
|
in your talk about deno you always say that you dislike centralized things like npm. But then why do you like crates.io alot? |
Beta Was this translation helpful? Give feedback.
-
|
Crates.io only impacts the building of Deno, not the usage of Deno. Seems like you are here to pick a fight. 😢 if you don't like Deno, just don't use it. |
Beta Was this translation helpful? Give feedback.
-
|
oh no, that was not my intention. I was just curious, I'm not here to pick a fight |
Beta Was this translation helpful? Give feedback.
-
|
@ry Hi Ryan, thank you very much for giving my slightly silly question such a great answer. You've actually sold me on the whole Rust idea. It's still a young language but I can definitely see how it cleans up the messy C++ ecosystem, and provides all the other Rust benefits like memory safety and a way cleaner syntax (not having 40 years of different legacy C++ syntaxes all bolted onto each other). I also appreciate how clean Rust's Traits are instead of C++'s classes. There are so many ways to shoot yourself in the foot or use wrong/outdated techniques in C++, so I can definitely see the safety enhancements gained by having Rust in Deno to ensure that random contributors have less ways to mess up the codebase. And perhaps it's also nice to not get 40 pages of template error messages (C++ errors) from the compiler anymore. ;-) When you, a person I highly respect, said that "I'll never start a new C++ project, because Rust solves it better", you've actually given me the final push to start learning Rust. I've been eyeing the language but always felt like it was too small and had too few libraries. But the fact that it's been the number 1 rated language on StackOverflow's "most loved languages" list every year for half a decade now, by a huge margin, says something about how good it is. I think that my worries about the future of Deno are going to be solved at the same rate that people are realizing how great Rust is and start creating/porting big libraries for it. Your message today, showing your support for Rust, contributed to helping me understand Rust's potential. And I recently saw news that they are starting to use Rust for the Linux Kernel. It would be great to live 20 years into the future, and having a huge Rust ecosystem already. But I think that things like rust-bindgen and existing Rust libraries will tide me over and have solutions for most things... So what I am saying is that I have decided to learn Rust and add myself to the list of users to help grow the language. It seems like it's starting to hit critical mass, when even the Linux Kernel decides that it's time to get rusty, and your words helped a lot. I don't think any other language in the world has more future potential than Rust, and I've kind of been itching for a reason to get involved. Thanks for explaining the rationale behind using Rust for Deno, and what it brings to the project. |
Beta Was this translation helpful? Give feedback.
-
|
Regarding GUI libraries: Rust has some fantastic up and coming libraries for GPU access (some written in pure Rust), most notably @crowlKats and @eliassjogreen are working on bindings to get WebGPU to render to a window. This should allow high performance, low level graphics with a modern API. If you care more about higher level GUI bindings (something like electron), also consider http://deno.land/x/webview. This binds your system provided HTML engine into Deno as a UI framework. |
Beta Was this translation helpful? Give feedback.
-
|
Wow, thank you for showing me that. That wgpu-rs library actually looks impressive. It's able to target WASM too. That would allow people to create custom 2D/3D GUIs, and that's definitely something an application developer would need. I had seen the webview library but got the impression that it looked like a small and possibly unstable project. I don't base that on anything other than few GitHub stars and never having heard of webview before, though. My main concern was actually the apparent lack of direct, native UI components. For example, libui-node or React Native. I wasn't able to find anything that could do that for Deno. And although a library like Electron/webview would suffice as an alternative, I just doubted the stability or maturity of webview. Perhaps I'm (hopefully) wrong about webview. |
Beta Was this translation helpful? Give feedback.
-
|
Remember Deno has only been stable for a year. I think more libraries with more features will develop over time as the ecosystem grows in size. |
Beta Was this translation helpful? Give feedback.
-
|
That's true. I like everything about Deno's design much more than Node's. And Rust is a fantastic language. I also love that Deno can be easily extended with Rust modules, which means things like UI libraries will be able to be tightly integrated in the future. And the V8 engine itself is brilliant (just-js was able to integrate V8 JS with the Linux syscalls with almost as low overhead as native C/C++, thanks to V8's fast JIT). Hmm. I think Deno's future is incredibly bright. |
Beta Was this translation helpful? Give feedback.
-
|
I realize that the Deno maintainers have no regrets, but what about the wider ecosystem of developers? You need to know Rust to contribute to Deno. C/C++ remains very, very active. And popular. For all the love Rust gets on Stack Overflow surveys, its hype is somewhat bigger than its size (not my words). It just doesn't seem fair to me that after more than a decade of developing Node.js core and extensions in C++, there's now no way for developers to contribute or extend Deno without learning a new language.
@ry This is more for curiosity's sake than any meaningful design decisions, but if C++ had a single tool for both pulling in dependencies and build system, would you be more inclined to (re)use it, or is it just the Rust syntax in general that makes it too good to give up? |
Beta Was this translation helpful? Give feedback.
-
|
Perhaps I should clarify what I meant. I am not asking/suggesting if Deno would be better off using C++. I am asking if there should be a way for C++ developers to use or extend Deno. Some examples (not comprehensive):
Please note that I have not expressed my personal opinions on Rust (or Deno for that matter). I just think that after all these years, developers who did work on Node deserve a bit more than "learn Rust or get lost." Shouldn't there be a way they can still contribute in some way? |
Beta Was this translation helpful? Give feedback.
-
|
Beta Was this translation helpful? Give feedback.
-
I don't think the Deno team owes anything to Node developers, other than perhaps gratitude for facilitating the demand for the type of project they're making. They definitely don't owe Node developers any compromises in their project design. Instead, Deno should allow people to use languages other than rust in plugins because it would be better for Deno. Saying Node developers "deserve" accommodation seems kind of entitled. |
Beta Was this translation helpful? Give feedback.
-
|
@Ruzie npm's problem is that it's a private company. |
Beta Was this translation helpful? Give feedback.
-
|
npm is owned by GitHub which is owned by Microsoft which is a publicly traded company. It is more appropriate to say that the npm registry is a closed platform with closed governance. It has been a desire by Deno to not force users to lock-in to a single closed ecosystem. npm in Deno is supported via a number of mechanisms currently and we are looking at ways of making it easier in the future. |
Beta Was this translation helpful? Give feedback.
-
Looking at the issues with lack of mobile ports due to Rust compilation issues (see other issues about that), and the lack of bindings for various GUI libraries since those libraries would require Rust ports/wrappers, makes me wonder something in general for @ry:
I know that Rust enforces safer code and memory management, but it seems to hinder this project from getting ports of popular C++ GUI libraries. Kind of the same issue that Rust in general has.
Beta Was this translation helpful? Give feedback.
All reactions