I would really love to see more interest in Rust from the embedded industry. With the influx of new developers into embedded programming I think there is an opportunity.
Yes, especially when Rust could be a relief to many pains embedded software suffers from. The counter argument is of course that embedded software is very entrenched. Often we would be happy if we could use a normal C++ compiler like gcc or clang instead of the fragile pieces of art that semiconductor manufacturers give us.
I think the embedded world needs a bit of outside blood. Being able to replace C/C++ with something a bit more conducive to helping people develop correct code would be a great plus. But the tooling also needs some help. Good tooling that isn't a nightmare to set up and maintain is crucial.
[edit: moved from another parent where it was less on-topic]
Considering how much of the actual programming in embedded isn't so much programming in C but metaprogramming in C preprocessor I wonder if 1:1 replacing this with another language/conditional compile pair would really be smartest move. It could be very interesting to embrace the fact that you are writing a program that writes a program instead of pretending that it's all one language. Don't hide the two-stage nature of embedded development, radically rip them apart. The first stage could be so much more than an afterthought if you elevate it from macro replacement to a type-heavy behemoth like, I don't know, a Scala DSL tailored for generating target-specific C/Rust/whatever source from a template. A template that interleaves precompilation and target language, like C source already does, but with a proper language for the first stage instead of a glorified find/replace sequence. This would have to happen in the form of cross-pollination from the web world because they have decades of learning from mistakes in markup generating code and almost half as much in dealing with the ergonomics of anything-to-js transpilers.
I suspect that if you'd separate all the customary compile time logistics into its own, distinct preparatory stage it would be much easier to get different languages from the ubiquitous "a compiler backend exists, but..." to actual pragmatic usability.
Rust has plenty of support for compile-time code generation of all kinds. It's generally viewed as something that should be done only as a last-resort, but it's there and fully supported.
I like how you describe it as "the two-stage nature of embedded development" because that is what in reality it really is.
In my opinion replacing the preprocessor with another code generator wouldn't be the best path forward though. I have worked with a bunch of code generators and it has always been a crutch.
I think something that ties in with the target language and is aware of the syntax of the target language, like Rust macros, has many advantages. At the same time I do not know enough about Rust macros yet to tell if they are really powerful enough to help here.
The problem of embedded isn't really a lack of implementation languages but rather that the ones that are being used are not very conducive to correctness and that the tooling and operating systems / SDKs / frameworks are awkward.
Much of embedded programming is about coping with quirky hardware and nudging bits and bytes around. And coping with rather sloppily put together tooling that requires a lot of hand-cranking.
Introducing higher level languages doesn't really solve this. If anything, it would make things worse because it would focus attention where the problem isn't. Adding lots of complexity by adding transpilers for a bunch of languages just makes things more complex and does nothing to solve the actual problems.
(Slightly orthogonal: just look at how horrid Zephyr becomes due to Python-based build tools that keep breaking things with every release.
Choosing Python actually made things worse because Python brings its own unmitigated problems. It screws up your day even before you get to do any programming)
One of the Rust-focused consultancies does a lot of embedded. There’s a whole rust + embedded specific conference now. It is a huge space, but some stuff is certainly being done!
Even C needs more love from the embedded industry, given how many still use C89 dialects, even if the tooling might already support something better, let alone languages that attempt to replace C.
C versions after 89 doesn't bring much to be honest.
So either you want more than what C can offer you, and usually the option is C++, or you're fine with what you have.
One roadblock is that many in the embedded software industry uses proprietary IDEs and tools like Keil, Atmel Studio, CCStudio and STM32CubeIDE. These propietary IDEs and tools are often based on Eclipse or Visual Studio Isolated Shell.
How good is the Rust support in Eclipse and Visual Studio?
Visual Studio should not be confused with Visual Studio Code. The latter has great Rust support.
I wouldn't say that they embrace them. I'd say they accept them. Often because this is what they are used to.
It is mostly newcomers to the embedded programming scene that go on about tooling because they're used to better tooling from non-embedded development.
I think one should pay attention to the newcomers and try to learn from them.
Does anyone know what the situation with Rust on the NRF is? It's all so new that there's no "getting started" guide, but even figuring out which repos to check out is hard.
There are crates for that [1]. But if you are new to rust and embedded, I would recommend to read through the rust embedded book [2] and practice with the stm32 board.
Ah, thanks. I'm not new to embedded, just to Rust on embedded. STM32s are a bit too overpowered for me, I usually use ESP8266s, that's why I thought the NRF might be a good bet (and better supported by Rust).
The compiler toolchain can cross compile to esp8266, but I think the pure Rust HAL is less developed there. If you use Rust to write components/code that link against the ESP-IDF, I’d imagine that would work?
I have the toolchain in a Docker container if you want to try it out.
I’ve been playing with Rust on the ESP32 using these tools on and off over the last couple of months or so, relying more in linking to the existing FreeRTOS/C HAL for accessing the peripherals.
One thing I haven’t quite figured out: a substantial number of onboard features are configurable at image build time, which then change the effective C libraries via preprocessor definitions. So, it doesn’t quite work to naively run bindgen on the header file once, and then embed that in a -sys crate and put it on crates.io
I think you can hand craft a -sys crate with all the appropriate config options as cargo features, but that seems both very time-consuming and likely to drift from the newest esp-idf versions. The alternative of running bindgen periodically (e.g. as part of the build process) works fine for one project, but then how do you publish reusable libraries online?
The ESP32 API surface is huge, and the libraries have many configuration settings. Rust-ifying this is a huge endeavour. One way could be to just interface with the vendor HAL using Rust wrapper, but doing that in a safe way that respects the lifetime rules can't be easy? Also AFAIK some lower level radio APIs are locked and hidden behind a binary blob, so the ESP32 can't be misused as a jammer.
Its very tempting to try this due to the huge amount of resources the hardware has, but it is a huge amount of work. Just being able to access the registers isn't very valuable, the ESP is about Wifi, BT and I2S. If there are no ergonomic Rust APIs for those, why even bother?
It's true and a bit surprising how much of the ESP32's API's are configurable or change based on config options. Espressif has built an impressive amount of functionality on top of FreeRTOS. It'd be nice to see Rust more widely adopted, but it seems to require an entirely rewritten SDK as so much of the API are C macros and defines.
Given that I decided to go a middle ground and use Nim with its new ARC (GC reference counting and move semantics). In theory it's less efficient than compile time Rust memory management but it works well in practice. Wrapping the C API is trivial and handles all of the C defines as well as writing in C, since it compiles down to it. And you can use any Nim or C libraries. It's surprisingly effective!
Except you still get data races from the C API's which Rust would likely help prevent. So it's cool to see someone toying with an xtensa llvm backend. I wish LLVM still fully supported a C backend so you could compile Rust to C and build that. Hopefully in the long run, more native rust libraries will exist for more full functionality embedded.
Was looking at Zig for the same reason, manual memory management but less error prone due to "defer", and a design ethos that is very reminiscent of C. Can even compile C directly! I really, really like the aesthetics of the language, but there is no xtensa backend yet.
Rust might just be a leap too big and can only work in blank-slate scenarios.
Zig seems promising too! But it’s still LLVM based (?), which seems to be a limitation in the embedded world. Nim’s templates/macros are nice for fiddly stuff. Not sure how Zig handles places where compile time macros can make things nice. Yah I figure in ~5 years Rust might be better positioned in embedded work as a goto solution.
Although, I'm more interested in learning STM32 development, due to recommendations that it is far superior to ESP32 boards. Anyone have some tips on how to get started on this front as a hobby?
I ordered one of the STM32 development boards when this was first published. It's fairly straightforwards to get basic blinky demos working, but it was about at this point that I ran out of steam, and you're basically left to fend for yourself, implementing anything more complex from scratch.
It may have changed since, but rust embedded definitely isn't hobbyist friendly unless the goal is to make things work, rather than to just make things.
47 comments
[ 2.9 ms ] story [ 108 ms ] threadConsidering how much of the actual programming in embedded isn't so much programming in C but metaprogramming in C preprocessor I wonder if 1:1 replacing this with another language/conditional compile pair would really be smartest move. It could be very interesting to embrace the fact that you are writing a program that writes a program instead of pretending that it's all one language. Don't hide the two-stage nature of embedded development, radically rip them apart. The first stage could be so much more than an afterthought if you elevate it from macro replacement to a type-heavy behemoth like, I don't know, a Scala DSL tailored for generating target-specific C/Rust/whatever source from a template. A template that interleaves precompilation and target language, like C source already does, but with a proper language for the first stage instead of a glorified find/replace sequence. This would have to happen in the form of cross-pollination from the web world because they have decades of learning from mistakes in markup generating code and almost half as much in dealing with the ergonomics of anything-to-js transpilers.
I suspect that if you'd separate all the customary compile time logistics into its own, distinct preparatory stage it would be much easier to get different languages from the ubiquitous "a compiler backend exists, but..." to actual pragmatic usability.
There are blog posts exploring some of the use cases for them, but it depends a lot on what you want to do.
I think something that ties in with the target language and is aware of the syntax of the target language, like Rust macros, has many advantages. At the same time I do not know enough about Rust macros yet to tell if they are really powerful enough to help here.
Much of embedded programming is about coping with quirky hardware and nudging bits and bytes around. And coping with rather sloppily put together tooling that requires a lot of hand-cranking.
Introducing higher level languages doesn't really solve this. If anything, it would make things worse because it would focus attention where the problem isn't. Adding lots of complexity by adding transpilers for a bunch of languages just makes things more complex and does nothing to solve the actual problems.
(Slightly orthogonal: just look at how horrid Zephyr becomes due to Python-based build tools that keep breaking things with every release. Choosing Python actually made things worse because Python brings its own unmitigated problems. It screws up your day even before you get to do any programming)
Whew. I’m glad it wasn’t just me that thought this. Zephyr devs seem really blind to this. And the swap to CMake didn’t help things.
https://blogs.itemis.com/en/eclipse-mita-bringing-software-d...
And yes, C++ would be a better option, provided that up to date C++ compilers exist at all.
Also declaring the index variable in a for loop like for(int i =0; i < 10; i++) is extremely helpful.
Those two alone are extremely beneficial.
How good is the Rust support in Eclipse and Visual Studio?
Visual Studio should not be confused with Visual Studio Code. The latter has great Rust support.
If you want their buy-in into other toolchains, you need to meet them where they are.
It is mostly newcomers to the embedded programming scene that go on about tooling because they're used to better tooling from non-embedded development.
I think one should pay attention to the newcomers and try to learn from them.
Also, "embedded" is a massive field with very different cultures and needs.
[1] https://github.com/nrf-rs/nrf-hal
[2] https://rust-embedded.github.io/book/
https://github.com/lupyuen/pinetime-rust-mynewt
https://medium.com/swlh/sneak-peek-of-pinetime-smart-watch-a...
I have the toolchain in a Docker container if you want to try it out.
One thing I haven’t quite figured out: a substantial number of onboard features are configurable at image build time, which then change the effective C libraries via preprocessor definitions. So, it doesn’t quite work to naively run bindgen on the header file once, and then embed that in a -sys crate and put it on crates.io
I think you can hand craft a -sys crate with all the appropriate config options as cargo features, but that seems both very time-consuming and likely to drift from the newest esp-idf versions. The alternative of running bindgen periodically (e.g. as part of the build process) works fine for one project, but then how do you publish reusable libraries online?
Its very tempting to try this due to the huge amount of resources the hardware has, but it is a huge amount of work. Just being able to access the registers isn't very valuable, the ESP is about Wifi, BT and I2S. If there are no ergonomic Rust APIs for those, why even bother?
Given that I decided to go a middle ground and use Nim with its new ARC (GC reference counting and move semantics). In theory it's less efficient than compile time Rust memory management but it works well in practice. Wrapping the C API is trivial and handles all of the C defines as well as writing in C, since it compiles down to it. And you can use any Nim or C libraries. It's surprisingly effective!
Except you still get data races from the C API's which Rust would likely help prevent. So it's cool to see someone toying with an xtensa llvm backend. I wish LLVM still fully supported a C backend so you could compile Rust to C and build that. Hopefully in the long run, more native rust libraries will exist for more full functionality embedded.
Rust might just be a leap too big and can only work in blank-slate scenarios.
"There is no hidden control flow, no hidden memory allocations, no preprocessor, and no macros."
Although, I'm more interested in learning STM32 development, due to recommendations that it is far superior to ESP32 boards. Anyone have some tips on how to get started on this front as a hobby?
It may have changed since, but rust embedded definitely isn't hobbyist friendly unless the goal is to make things work, rather than to just make things.