Microsoft Is Abandoning C# for Rust! Now What?
2024 ж. 4 Ақп.
87 434 Рет қаралды
Check out my courses on Dometrain: dometrain.com
Become a Patreon and get special perks: / nickchapsas
Hello, everybody. I'm Nick, and in this video, I will discuss Microsoft's decision to migrate from C# to Rust for some of the performance-critical services in the Microsoft Office 365 product.
The Reddit post: / came_across_a_job_post...
The job post: jobs.careers.microsoft.com/gl...
Workshops: bit.ly/nickworkshops
Don't forget to comment, like and subscribe :)
Social Media:
Follow me on GitHub: github.com/Elfocrash
Follow me on Twitter: / nickchapsas
Connect on LinkedIn: / nick-chapsas
Keep coding merch: keepcoding.shop
#csharp #dotnet
Please create Rust for C# developers intro video
just look for the Rust book
Please don't
@@ripple123 The Rust Book is a really amazing starting point. It's really well written and easy to understand. I have read it entirely, as well as Rust by Example and the Rust Async Book. Rust is an amazing world.
I'd watch it...
Yeah, a little introduction video to see the main differences between rust and C#... Why not🎉
I'm a software engineer at msft and this Rust hype is pretty new among O365 teams, from the teams I know (mine included), this "transition" is much more of a experimentation than anything else, just a small group of people working on some small scope service rewrites to see what are the performance gains and so on, Microsoft is definitely not abandoning c# lol
Absolutely. There is no sense in rewriting a single line of production code without proper business justification. Rewriting small and controlled use cass would be able to build constructive arguments to either go for a migration, migrate only critical parts, or do nothing at all for default cases.
That pause in “keep coding… in C#” had me chuckling. 😂 By all means, I was already interested in learning Rust (as a C# dev).
(as a millinonaire)
@@RaZziaN1 glad someone got the reference. 😂
There are 2 types of Programming Languages: Those that been hated and Those no one used. Rust is both 😂
I'd be super interested in a Rust intro video
Me too
Go watch primeagen for rust content 😊
Yeas please
google nicholas matsakis and aaron turon rust programming language, great intro book with examples and questions
I would also be interested in more advanced rust videos tailored to C# devs. Most videos I came across were either for absolute beginners or for veteran C/C++ low level engineers.
Nick: Microsoft isn't really abandoning C#. Also Nick: Let me use that as a clickbait title for the youtube algorithm.
Was gonna say that. 0:18 "Not a big deal, but I'm gonna make a worried face on my thumbnail anyway".
The games we must play in the world we've built.
Hate the game, not the player
You wouldn't be here if you haven't seen that title, wouldn't you?! 🤨😏
I'm currently picking C# for the core of the app, leveraging on class management, and rust for my computation library. It feels like a winning combo right now.
It works really very well
Doing the same thing for a game simulator.
@@jeffmccloud905 no benchmark yet. It's more philosophical than rational. I actually don't even know if I can save some memory state going back and forth from C#. Oh, and I am testing functions from Jupyter in vs code...
Can you afford on developing application in 2 different languages? How much will you get?
@@ProstoDoCelu316 I am single pseudo dev, coming from finance. I'd reconsider if I was in a team with a lot of pressure. But this stack allows me to really separate concerns. I do more experimentation than testing in rust/jupyter (matrix style computation in stats). I actually do the compile from the jupyter notebook. I will be able to reuse the library as a consultant if I need it, I think it is easier to call rust lib than managing .net lib (which I experimented a lot in matlab) It's super simple in python: !cd E:\Dev\MyFolder && cargo build import ctypes import _ctypes from ctypes import CDLL tgt=r"E:\Dev\MyFolder\target\debug\dllName.dll" lib = ctypes.windll.LoadLibrary(tgt) y=2 x=3 z=lib.add_numbers(x,y) print (z) # release the library to be able to recompile on th fly _ctypes.FreeLibrary(lib._handle)
A suggestion would be to make a series learning Rust from a C# developer perspective. It'll be great, because we can see your learning methods and how you approach known problems with unknown tools
If you read more comments on that post and others talking about the same thing, you'll see that the team stated that the reason they picked rust was because it was rewriting a small set of existing c++ implementations that had to stay native and rust was the most logical choice to do that.
As opposed to just keeping them as c++?
@@robwalker4653I believe they're expanding them also but yeah more teams there are pushing conversion of c++ to rust as a lot of other platforms are. Reasons I've seen are difficulty of finding c++ devs and rust has compile time memory safety without needing a garbage collector.
@@jbarker83 there are plenty of C++ devs, but having that compile time memory safety is the killer feature that people want.
@@evancombs5159 C++ is also full of footguns. I'm primarily a C++ dev, so I know how to avoid most of them, but they're still there, and even experienced developers run into them occasionally. For example, if you call an overloaded function that has std::string and bool overloads, with a string literal, e.g. "hello world", the bool overload will be called, because const char* is a pointer, and pointers are implicitly convertible to bool, and this takes precedence over the implicit constructor of std::string. I want to learn Rust, too.
@@evancombs5159 Google are also now heavily into Rust, so far in Android. They've published stats on their reduction in security vulnerabilities. They're also looking at Chromium but that hasn't really got going yet.
This is really validating actually. I started teaching myself C# to build a lab inventory management system at a small biotech company where I started out answering phones, and I mostly picked it just because I was able to find resources that helped me picture what all I'd need to learn and do early on. It's worked out pretty well, they've been using my system for months now without issue and I'm continuing to learn everything I can to improve the design, but it was kind of a random choice to start with. But a language that does everything well enough, so you don't need a huge team--yeah, that's exactly what I needed, what luck I picked so well.
Love both languages. We use: - Linux (including workstations) - C# Blazor (web) - C# MAUI (mobile) - Rust (all else + backend) - postgreSQL (sharded) Devtools: - jetbrains subscription + tools - self hosted git, youtrack, hub, team city - hashicorp tools (terraform, nomad, packer…) - self hosted wg vpn services This works well for us and I wouldn’t really change much if anything in the stack The Blazor side of things gives us the ability to be server side or WASM (or a mix with 8.x) where this level of flexibility is important to us
The great thing about C# (and .NET) is how comfy it is. Sure, there are faster languages, but C# hits the sweet spot of comfort and performance. And MS makes it faster and faster every year, too.
Which ones are faster?
@@bikeindustryshill C# and Rust (and C++, and etc) have comparable speeds. Rust a lot faster in specific scenarios, and can be easier to make programs run a lot faster than it would be in C#. But if you are writting general software code, C# will not be more than 15-30% slower than native languages, in some very specific and rare cases it can be even faster due to JITting.
If you need to ask that question, it doesn't matter; use what works for you, your abilities, and your company.
@@diadetediotedio6918 Thanks!
@@andersborum9267 I wasn’t going to change what I’m using for my company, just asking out of interest
I remember when NodeJS was first announced proudly claiming that it's able to handle 10k requests. MS uses every tool available. Until they announce end of support for C#, it will be fine.
I think Rust and C# just have different areas of application. You can do anything in Rust of course. But considering the learning curve, companies will still prefer simpler(means cheaper in many ways) tools in every scenario possible. So there is no threat to C#. In my opinion C# is now a golden standard for any general business use case. And you go Rust, Python, whatever if you know why it will be better. I think that as a person who does every personal project in Rust now, but works full-time as a C# dev.
I was doing PHP with the Symfony framework 8-9 years ago. The service could achieve about 300 requests/sec on a small VPS, but it was a small startup, so it was good enough.
We have a tiny home automation product line where rust is used to write the controllers for the devices not because its "fast" but because we can cram a lot of functionality into very low powered device that way. C# is still used to create the user facing contoller. Was it a good idea? Who knows but our small team likes it
Stick civetweb on your controller and use the web to create a ui for it. Works surprisingly well.
As a developer who writes C# and Rust (and others), I'll say the first step in moving from the former to the latter is "forget everything you think you know." It's a lot easier that way. I'm way more productive in C#, but Rust has niche benefits for specific scenarios. Neither one is going to replace the other.
Other points aside, Rust has been at the forefront of WASM and WASI for a some time. It makes sense that critical core components would be written in Rust and, eventually, directly consumable through WASM and, in the not too distant future, WASI. WASI is effectively containers without the container (and faster). Microsoft has already announced plans to support WASI with .NET. I won't be surprised if we don't see an experimental or preview release in the .NET 9 timeframe and the first release by .NET 10 or shortly thereafter.
Don't think there is a real competition between C# and Rust... As there is no competition between Python and Rust... The right tool for the right job.
There is competition on everything, but also there is not, because tools can be used together, so most of the time I don't think these battles people do in internet are worth anything.
I've been learning Rust for a bit - it's indeed such a beautiful language! I now really miss Result, Option, redefining variables, enums, from C#, but at the same time, C# is more aesthetically pleasing to me, while also being really flexible. If only there was a hybrid of these two, it would be a perfect language, haha
You can download language-ext package and have these and then some.
Crust
Apparently you can implement Result, Option in c#
@@johnnykeems2911 of course you can, I even do it in my projects sometimes, it's just that they're not integrated in the whole ecosystem
F#ck, why can't there be a language that combines the best of both?
I would be interested in a video on the use cases where a mixture of RUST and C# would be useful.
I love both C# and Rust, I don't see most C# devs getting over the frustration hump of Rust. Nor do they need to for most situations. There is a Rust for C# book online, I would recommend that for anyone interested, you too Nick,
There is more than one frustration hump, to boot.
Is that Seb Nilsson "From C# to Rust?"
I wish MS would implement Rust code embedding within C# code, much like how TurboPascal could embed inline x86 assembly code. I know, I know, that's not how Interop should work, but I'm sure the MS C# magicians could work out emitting IL and Rust compilations behind the scenes? To really go into fantasy land, imagine passing Linq constructs into Rust for tight loops etc.
I've written a little Rust, but a tutorial from you would be great!
Title = Microsoft Is abandoning C# for Rust! "Yup, you read the title right, it is not click bait" 15 seconds later: "People are trying to sell it as Microsoft is abandoning C# while that isn't the case"
i would like to see introduction to rust from perspective of c# developer, what are downsides and what would actually make someone to develop software in rust instead of c#
Downsides are longer compile times, steeper learning curve, less jobs, and smaller ecosystem. But the upsides far outweigh the downsides imo.
It is funny to learn something new and exciting as well
Here's one reason: Rust doesn't have nulls, so you will never have to deal with hunting down the cause of a null reference exception or null propagation ever again.
@@MarshallRosenstein How does it deal with empty boxes? And how do you check if something is constructed or not. It must have some mechanism for representing non-existence? Null is a real thing, it has consequences. Are you saying Rust doesn't allow consequences for something being missing?
@@gavinw77 That takes a somewhat long explanation. Rust doesn't have constructors, but the equivalent of static factory methods only. The compiler forces you to set a value for every property of a struct when you create an instance (much like a c# struct, which requires every field to be initialized - although in c#, structs are value types), or define defaults. To represent non-existence, functions / methods return a well-known enum called "Option" which can have a value of None (which would represent missing) or Some (which represents an actual thing). Unlike C# enums, Rust enums can have an object payload associated with each arm, so when you get a Some, you can extract its payload. The compiler forces you to deal with both possibilities (Some or None) or the code fails to compile, so there are no unexpected missing objects at runtime. In C#, if you don't handle all branches of an enum in a switch statement, you just get a warning in the IDE. A more general way to think of it is that, regarding null(ish) handling, the Rust compiler treats as errors what the Visual Studio analyzer just treats as warnings.
Reminds me of the saying "Make it work, make it right, make it fast". Steve Smith (Ardalis) did a great blog post or podcast about how you only need to "Make it Fast" when that becomes necessary. You're not going to invest development time and energy in making a service fast when you don't know the capacity requirements of that service. Just like you wouldn't bother with "Make it right" for a proof of concept / throw away prototype. Once you know the requirements, and if the requirements are 500,000+ requests per minute and not spinning up server instances like there's no tomorrow, then you look at re-writing in a language that's going to take 10x as long and 10x as many heads.
It will be great to hear how to integrate Rust code with an existing C# code in-memory without remote calls, e.g. to re-implement hot-parts.
To be honest after years have found rust that I want to learn to grow as a developer specially to leverage and use the power of highly native code that languages like C,C++ and rust provide. And out of them all at initial steps rust seems reasonable
Ive been coding c# for years, but now trying Rust so by all means do a video (I really like your videos)
I’ve bought a ton of courses from DomeTrain and would love to buy an intro course to rust. Hell I’d even buy one for Go too.
i'm very interested on rust language, video about it will be nice, but not stop on c# :)
Nick, I like and watch your videos because you present important topics, explain them well, and get right to the point. For those reasons and because I'm interested, I'd like you to make an introductory video on Rust.
Yeah totally valid approach. If something is slow and you can't get it much faster with the environment (language) you are using, rewritting it in a low level language can drastically increase performance - especially when you use features like SIMD directly. I am always interested in deep level programming, so it would be nice to get a introduction to rust - for right now, i am totally fine with C99 and C#.
Checked out the borrow checker etc. But ultimately I concluded that is an easier way to write c/c++ level programs. So if you didn't need to write c++ before then you don't need rust.
Yeah Rust is basically the "replacement" for c/c++. But i still think that it makes sense to use Rust in a lot of cases where you wouldn't have wanted to use c++ before. C++ has just a higher chance to shoot yourself in the foot. Memory safety, safety in general and so on. Rust was basically designed to fix the weaknesses of c/c++. And it has a lot of convenient features, e.g. async and stuff like that which would be a pain to implement in C++. And a really strong point is the ecosystem, Rust comes with basically everything you need to start a production-level project while in C++, you first need to decide from a ton of compilers, build systems, package managers (if you even use one) and so on. In summary, i see why people wouldn't want to write some service in c++ just because they don't want all the headache, but would still choose rust now over a higher-level language.
Rust introduction video for C# developers is a video iv been searching for!!!
Just start with a generals introduction - C#/OOP mindset will not help and few on the language level translates into Rust - there are no exceptions but you get borrowing/lifetimes which does not really have a corresponding idea in C#.
as a Rust dev, I'm excited for more institutional support for this lovely language :). I've always said that Rust is great for writing the building blocks, and a garbage collected language (C#, Haskell) is great for putting them together. This doesn't have to mean the end of C#, but you can bet your ass those cpu bottlenecks and memory freezes will be a thing of the past
Basically I only learn C# to code a windows 11 native app. For web services I'd use JS / Rust depending on many factors, and I guess I'd use Flutter for multi platform
I worked in Prime Video and during premier league we had scaled a couple of services up to support 800K requests/s. Services were in Java. I doubt the reason is just to save milliseconds. It's most likely some low-level parts in C++ that are integrated into the C# implementation that they want to refactor.
They are just rewriting it to Rust where they would have chosen C++ otherwise in the past. If they did that instead now, it would not even reach the news. Actually I would also really not be surprised if C++ parts of the CoreCLR also end up in Rust. Although I can also see a possible C# future there as well with the evolving NativeAOT basis. Both of these can be an extremely good choice. I am now mastering both languages and I love it. I have written C# and lots of other code for 20 years, but I a year ago I also started with Rust and I do not regret it.
I think that for anything that is not super critical performance wise C# is probably better, I am pretty sure development is faster and MS definitely have more C# developers at hand than rust developers. But for those really hot paths, rust can be the best choice. Even with all the performance improvements in C#, since you still have the GC, memory management cannot be fully predictable. With rust you can much easier know how it will behave and can ensure that the container has just the perfect amount if memory available. And when your talking tens of thousands of instances, that adds up quite a lot.
@@davidmartensson273 I agree, but it is also good to see that for a GC language C# has quite advanced options to minimize GC load nowadays. They have improved and extended a ton of things around value types and stack usage. Just think about all the struct types, spans and the new internal arrays. Together with the ahead of time compilation, it looks also really interesting. No chance to win from Rust or C, but dotnet is already faster than Go now. Like you say, it's also about developers. But I think also a bit about languages being written in itself as a beautiful thing. Already true with the compiler, but not fully with the runtime.
@@jongeduard Sure, and if you use all the latest performance features and can limit GC overhead, I would not be surprised if you can at least get on par with Java, if not better, BUT it is also about where you put the time. For most projects, spending time on your logic will give more gains :) We recently switched some older netframework code to a new NET7 project and got 25 times faster responses. Almost all of that was due to better logic, and better caching, and we can probably improve caching a bit more if we ever need to.
@@davidmartensson273 I thought Java was beaten anyway. Java still lacks proper value types except the primitive ones. Even from my own testing with Java, which I rarely touch these days, I have never seen it winning from DotNet. What about your own projects, true. I was really talking in relation to what they can do with the runtime and optimizing that. When working with IO, network and databases stuff your first attention goes to async code. And that does not even work together with stackalloc, references and all that, since async code cannot even maintain a persistent stack. Same thing with LINQ and iterators. And I actually prefer that over the low level stuff normally, because functional programming features are great.
@@jongeduard I think that since the Java runtime have been around longer and they did focus on performance early on they have done a lot of work optimizing for many more specific use cases that might help on these performance tests. But for more generic code that does not match some of the selected cases, C# could very well provide better performance. Also, some of the big performance increases in C# has come the last few years and I do not know how actively the aspnet example is optimized for the workload in the test. Especially Net7 provided a lot of low level improvements that might not have made it into the test suite they are running. I have not to much personal experience writing for java ;)
It would be great if you could explain why C# is here and there slower than Rust. Can C# improve for becoming that fast? Thank you Nick 🙂
C# still uses a garbage collector, that will always have some performance overhead over directly controlled release of memory. Another part is all the checks to prevent misuse of memory and objects and the reason C# have the unsafe keyword that you can use to bypass some of that for better performance. And finally, rust is a more modern language where mozilla designed it to enable lots of performance optimizations by the compiler, slow datatypes or constructs was never added to the language to begin with. C# originally was a replacement for visual basic and in the early days they openly admitted that performance was not the top priority but ease of use was. C# have since evolved and with spans and many other improvements it can compete with a lot of languages out there, especially for the average developer. But for C# to really compete with rust, MS would have to start removing things from the language that makes all the extra checks necessary and that would mean major breaking changes beyond anything most users would accept. Also, C# is mostly run on a runtime. Sure it has JIT compilation and now they are adding AOT that also can speed things up, but since all of that is added after the language was designed, they just cannot optimize as hard as rust can. One of the benefits of the limitations in rust is that just like Go, it can parallelize things aggressively in ways C# just cannot do. You can probably find some special cases where C#'s new features can compete with the equivalent rust code, but for any serious application, rust will be faster. Maybe not too much in some cases but I doubt C# would be really faster in any example, at best they would end up on par since they both make optimal use of the platform they are running on.
Rust is great, and C# as well. You can use both at the same time and get most of the benefits of both.
That is something I'd like to try one day. But I'm not even keeping up with the performance features of C# let alone integrating another language.
How about working on a project that combine both c# and rust. Would it be possible for a certain project?
The Seq log server is written in C# and Rust. they discuss it in their blog posts. The company is Datalust.
I would like the rust introduction for C# devs
being a lot busy and with tons of thing to learn around (still learning after 20+ years of dev, you never stop learning) having a C#-to-rust intermediate/advanced video would be really appreciated, that's actually what stopped me from learning it from scratch
Definitely interested in the Rust for C# developers video!
You don’t “automatically” get that 40% performance. Getting peak performance in Rust (or any other language really) is not easy…
Also techempower benchmarks are bullsh*t. just-js clearly showed that.
Can't they just use UNSAFE and unamanged code in c# where performance is so critical, instead of rewriting whole projects to Rust?
It is also very similar to why so many learn React because of Facebook. It causes excitement and excess demand that then make companies invest in it just because "it is cool" and devs are available. Now, React did solve a lot of problems. But there are other frameworks that are at least as good. But now the companies have gone all in on React. Keeping the demand high to maintain that code.
I prefer vue3 honestly.
Further hardening + extra performance.. What's the down-side?
I am working on a crowd simulator for Unreal Engine (for far too long) and I needed a good environment for prototyping. I have tried with Unreal but it comes up to consume a lot of time killing any hope. Next I moved to Unity and c# with it's DOTS, but damn, it has too many crap I should knew about to justify it as a good prototyping solution (and I need ECS anyway). And then I have tried Bevy engine with rust and finally it feels like prototyping. Types casting was a problem for me, but now kind of deal with it. Overall it is just enough for my purpose. There a bunch of crates (libs) that is easy to integrate (too easy after years of unity and unreal) and so I don't needed to write stuf... I already made in other languages (too many times). So yeah, it is not like you starting from scratch with broken arms waiting them to heal while still making some code (as usually I do). I almost not feeling frustated as hell and lost whiile working with it.
I think learning Rust in general is a great step to being more software developer, or actually software engineer, than just being a coder or a backend or frontend thing guy... Also I have been struggling to find any remote job lately, so guess it is the step that i might try my luck into.
In fact, I am from the Czech republic :-)
Services evolve, it is not surprising, the service discussed might have evolved to point where original creators did not foresee. There is a right tool for every job.
I think that Rust is beneficial to understand, it forces you to rethink a lot of bad habits.
Definitely interested in Rust and how a c# developer can port his skills to it.
Can someone tell me what Nick is saying at 0:30? "If you like (?) content and you want to see more" I can never understand it 😅
bro starts rapping
"that type of" I believe
you should watch it in x1.50
I thought it was C and C++, now it's C#? I think it's a good thing, C# isn't going anywhere anytime soon. I was going to learn Rust anyway because it looks like a fun language.
It is a fun language, the compiler can be annoying sometimes, but once you get it you get it, and it stops being a problem at all, just take it easy.
@@diadetediotedio6918 And the compiler messages are very good. One of its killer features.
RUST is overrated! Mostly people who didn't write anything big on rust overidealialize it. When you write something complex on it, trust me, it becomes full off sh*t.
That's true for almost any low-level systems language. I've worked with modifications to audio chip device drivers with custom Android (Linux) Kernel for old Samsung phones. You'll see lot's of "hacks" and weird stuff that's written because it takes HW "sympathy" (memory layout, cache coherence and usage patterns etc.) very seriously. Written in C of course.
Nah, I already made some relatively complex projects on it (and I'm making one right now), it is pretty good and scalable.
Honestly I'd be up for a Rust introductory video.
Hi! I had a look on Rust, but unfortunately had no time yet to deep-dive in it. A cool language, and you can find some very similar concepts in the latest c#. But, I have a hunch that they need rust developers because of python integration in Office. And integrate rust into python is super easy. Just, if you have time and you do not mind, can you ask around your Microsoft friends? Ps.: Yes, I would love to see a good video about rust.
I have been using C# for 20 years and Rust for 3 years. Rust is a great language with some great features. It has no nulls, it uses the results pattern instead of exceptions, etc, etc But what people tend to miss is that Rust is a paradigm shift. By tightening slightly the rules of what is allowable code the Rust compiler is able to do so much more for you. It achieves at compile time what C# does at runtime with it's garbage collector for example; it is able to verify that there are no race conditions at compile time etc, etc. This improves performance AND productivity. I find it very difficult to see a case for C# beyond Rust is hard to learn ... and it is hard to learn, maybe too hard ... time will tell.
I love this - we probably get some really great MS specific libraries for Rust this way
Send this job post to the primegen 😉
I can understand why Rust instead of C# + JIT, but what about C# + AOT ?
Wondering if Rust will be a passing fad like Ruby on Rails was 10 years ago.
Well, we can never predict the future but I'd say it's unlikely unless another language can do what Rust does and do it better. Bear in mind that most of Big Tech is now using Rust and are members of the Rust Foundation and are using Rust in production - Microsoft, Google, Amazon, Meta. Apple have been quiet but I bet they are using it too.
I really don't get that benchmark website. Java is higher than C# even though C# is faster. Even C++ is lower than Java.
C# as a language is faster, but Java runtime is still more optimized (having a number of years head start). C# is catching up piece by piece but especially in these types of benchmarks they are looking at examples that really need massive performance, and that means that the closer C# gets, the harder the next gain is going to be. And also, java can be fast if you write it the right way, its just that some libraries and constructs in java are badly designed from a performance perspective giving it a bad reputation. Just like that you CAN write safe secure and stable c and c++ code, but its also easy to write a pile of safety violation garbage if you do not know what your doing. Minecraft is java (at least the original java edition), and that is good performance, but is also uses some quite fast libraries avoiding much of the easy to use but slow ones.
@@davidmartensson273 does it mean that in general Java based apps are faster than C#, because they run on better environments?
I dunno. I guess it's good to improve performance. But Microsoft has a lot of C (and C++?) code that could be rewritten in Rust to improve security. I would have considered that a higher priority. Still, could be worse. At least they're not rewriting memory-safe code in an unsafe language. (Well, apart from Minecraft, but that's just a game.)
I study rust a little, but because I want to. I like to think it even made me a better dev in general, but, I don't expect to get a job with it. Even if I lose my job, I know the chances of landing a C# job are higher than rust, even though I haven't worked with C# or any serious backend in almost 5 years
There are jobs out there but the entry barrier is higher. E.g., you need the domain experience as well, e.g., blockchain.
"You heard it right! The title is not click bait!" - Goes on to explain how the title is click bait
Sadly I will never can use Rust as efficient as C# for game development purpose, try for example create UI library for your game engine with hierarchy, events, etc in Rust... I know you have to think differently when writing code in Rust but it just create problems and wasting huge amount of time. On top of that - compilation times are longer, no hot reload, no reflection :)
Maybe you could use something already made, like bevy. But Rust has a different feeling than C#, if you want to be productive on it you need to change the way you structure your programs. But listen, I think if you are making a commercial project and not just for learning purposes, C# is still amazing as it is for game development, and as you are more familiarized with it it will be a lot faster do develop as well, so don't worry trying to convert everything to Rust. And there is another bonus on it, when you really need to accelerate your code for something very specific, nothing will block you from using Rust from C#, because C# can FFI with Rust, so you can get the best of both worlds.
Well, that's subjective. After a few years in Rust, I perceive it as a more logical approach than the classical OOP. I have no problems writing "game engine with hierarchy, events, etc" either in C# or Rust. I.e. you only think that the classical OOP is easier simply because you are used to it. When you know both, you realize that they are just sides of the same coin and you can achieve the same goals in either of them equally. But it is not really a Rust exclusive concept. You can adopt that paradigm in other languages as well, including C#. Simply start to use composition and interfaces (default interface methods added in C# 8 effectively allow you to make traits) instead of inheritance and abstract classes.
@@diadetediotedio6918 yeah, C# with Rust can be very powerful combo I think!
SImilar for me, I will have a look on rust, but having used C# since version 1, how many years will it take me to be good enough in Rust to beat what I can do in C# and how much delayed would any project get due to that? Probably to much time. If I had the money to hire ace rust developers to build some parts, sure I would consider that, but I would never be able to justify the cost with the scale of the project ;)
It's not good for game dev because prototyping and quickly iterating is slow. Another reason it's not good for game dev is that the extra effort you have to spend to pass the compiler's correctness check is not worth it for games. (there's a lot of correct code that doesn't pass the check)
Introduction to RUST would be great!
People tend to overreact so much sometimes
I have to look at using Rust as there is no C# framework to create smart contracts. The future is web 3. technology An introduction to Rust, including comparing C# equivalent would be appreciated.
I've using rust for 8 months and i made bunch of liblaries, a cli app and 3 websites(which one of them is currently deployed). It's the best language i've ever seen, because there is no runtime errors. That especially makes it awesome.
I'm just trying to write backend on rust for my pet project. I have 10+ years .net expirience. And in rust - it's dificalt to do some thinks (may be in .net it would be same from 0 knowledge). So I still trying but from time to time I'd like give up
In my first application, coming from C#, I got stuck for 3 days at one point, doing something that would have been dead easy in C#. I never did solve the problem, so I approached it differently instead. Also I initially tried to replicate the full C#, which had interfaces. But that was too hard so I just stuck with concrete structs until I got something working. Then months later after I'd learnt more I was able to match my original intent. It require patience and practice.
There was no need for the click bait title. Your content is good enough to pull people to your channel without cheap tactics like that. :-)
Sink or swim. KZhead doesn't care. You need inflammatory, scandalous titles to get on the radar
I just wonder why Rust, why not modern C++, like C++20, C++23, with all those beautiful things it has and very powerful optimizing compiler?...
I guess using C++ would be a downgrade at this point - you just give up on memory safety which is what C# gives you thanks to a GC. And Rust compiler generates pretty fast code, with perf on par with C and C++ though there are plenty of possible optimization that can be implemented.
@@cysia3683 Hmm, but Rust is also not a "GC" language, it just has a few specific features for memory management, that let you prevent some most common errors. But the same can be done with modern C++. If you know how :)) That may be the point. Maybe Rust is just easier to learn.
@@user-qw3uu8gm8z I meant C# achieves memory safety by having a GC and Rust can do without it. And modern C++ doesn't solve anything. While it may be harder to accidentally leak memory or cause an use-after-free you can still easily get an iterator invalidation for example. I feel like in simple cases modern C++ helps writing safer code, but in less straightforward scenarios you get even bigger problems. Each added feature has some weird edge case that introduces new ways to shoot yourself in the foot. Sure, you can use all the available static analyzers to catch some of the issues, but it is less reliable than using a language that is designed from the ground up to help writing correct code. And there is nothing in C++ that can help in preventing thread safety issues.
@@user-qw3uu8gm8zI was referring to Rust's borrow checker as a way to avoid the necessity of having a GC. With C++ you don't have a GC, but you lose memory safety that they already have with C#. And modern C++ doesn't solve issues that Rust is meant to prevent. Sure, you have smart pointers that help to avoid accidental memory leaks or use-after-frees, but other issue like iterator invalidation or thread safety are still there. Which sanitizer for C++ can replace the static analysis in the rust compiler? It feels like when C++ introduces new features to the language it maybe helps with some easy cases, but in more complicated cases it gives you another edge case that you should be aware of unless you want to be surprised later.
I tried doing a bit of Rust purely for interest reasons but real life got in the way. I'd love videos on it. I really want to learn a language that's a bit lower level but I really don't like C or C++. I've considered Go, but Rust just feels like the right thing to learn out of the two.
Microsoft Learn has a Rust course, which I did. But I can't remember specifically whether it made comparisons with C# along the way. I didn't start with that course but with the online book - which everyone has to read the bulk of at some point. You certainly need to read the first 10 chapters.
This looks like mixing apples and oragnes (Why only h2o, then below we can see DBMS names like postgre?), and 'Rust' faster than 'C'. Nooo. However, it's nice to see Java blowing off C# and .Net.
We had this same problem with Silverlight/WPF vs Forms. Is MSFT supporting WPF, what’s the forward strategy if Surface doesn’t catch on , why bother coding in XAML if it is going away? Then little by little the community dried up. I don’t see this happening anytime soon with C#, but it does create unnecessary anxiety.
C# is so large that it would take decades, especially since more and more is open source. Silverlight was killed because the users had to install something to get it running and with both java and especially flash giving such plugins a bad reputation that became harder and harder while pure JS got better and better and mostly offers all the same features. C#, if you run it server side does not require any support by the client and with AOT, you would not need to install anything extra anyway. But even so, MS has been writing more and more of their own application using C# for the last 20+ years, they will not throw that away anymore than they will abandon C++ anytime soon, sure they rewrite parts in Rust and give it a decade and maybe C++ might be on the verge to be replaced, but they started that process when c# came, rust is just replacing the parts that c# could not.
To be honest Microsoft is the biggest problem c# has right now. It's the best language to use by far but the Microsoft stigma follows it everywhere. But people need to contribute more to the community for XAML for example there are 3 brilliant open-source projects(Avalonia, Uno, opensilver) but people still want Microsoft to do the same thing as they do. We need more companies to embrace c# or f# and not be a Microsoft only thing. And this rust nonsense will not end well with the rewrites if they wanted high performance they would have used c++ because it's better in every way then rust. This is just buzz word programming that usually doesn't work out in the end.
@@FilipCordas Rust comes with some very good benefits over c++ with safe memory management and pointer handling that more or less eliminate most of the common security bugs that plague c and c++. And its also better when it comes to automatically parallelism. Sure you can do the same in c++ but it takes a much more skilled developer to pull it off. Linux is also looking at incorporating some parts written in rust for all the same reasons. As for ms being c#’s biggest problem, I think that is only in some circuits, among corporations I do not think its a problem generally. Or where it is, its because they have invested in something else, oracle snd java, google and go. And especially for google, go has a different design goal that more closely matches their needs.
I did not see it in the comments but there are 2 things that are questionable. 1. Rust as an application language, you guys have to understand that people don't choose rust because of it's short and neat syntax (it's bloated), they choose it because it has an amazing compiler. 2. Why doesn't microsoft seriously focus more on performance? .NET 5 had some insane ups, but they can put a lot more effort into the compiler, instead of language keyword level of optimizations.
Those people are just overexaggerating to generate more clicks/views. Just like you did here by the way ;) In general it's sad to see that that trick still seems to work in 2024.
Do you remember when people said "Why to use Blazor, even Microsoft don't use it". And now this... I think there is a communication mistake from Microsoft. PS: About the 60% performance gain, everyone knows that this is the ideal scenario and never happens. But im wondering... if they can afford 10x devs/time why don't simply increase the scaling?
As with any large company, Microsoft uses the tech stacks that make sense for a particular use case. For most use cases, then yes, something like C# is a great solution, but when performance is critical or you need a lower level system language, Rust is a great option and a safer alternative to what used to be the default go to languages when you needed performance in the past (C and C++). It wouldn't surprise me if some of the Windows kernel and drivers were moved to Rust in the future as well.
One thing to keep in mind is that while well written rust is going to be faster than corresponding C#, it also depends a lot on what the code actually does. If to much of the work is waiting for databases or other external services your main benefit would be to be able to handle more waiting in one instance. Also, while you might save some hardware, if you need to spend time learning rust or hiring a rust developer and then build it, unless you are running hundreds or thousands of instances, the saving of 40 % of them will not make up for the cost of rewriting, and if your rust expert leaves, finding a replacement is going to be harder than to find a new c# developer, at least for the next decade or so. For really big sites like MS and others, sure they will save tons on it, but for most companies I would see it as more likely that if you have developers that have learned rust already, sure look at using it for some new development to test it out. Rewriting something from the ground up even in the same language is lots of work to begin with, adding in a new language and possibly a new team will add to that. That said, Rust is interesting and as a replacement for C or C++ I think its very good.
"It wouldn't surprise me if some of the Windows kernel and drivers were moved to Rust in the future as well" It's already happening. They've written about it.
And finally, C# needs a good escape analysis system, or at least a way to allocate classes on stack unsafely, oh the things I would be able to do...
If you need to work on stack use structs and for collections you can use inline arrays/buffers/Array pool
@@BasyrasCZ I already use them, I'm saying it would be benefitial to be able to allocate classes on stack or at least have a custom management over where the classes would be stored in memory, so things like bump allocator for classes would be possible. Currently, all C# classes are indirect pointers in arrays and in stack, so it has a cost that we can avoid in specific contexts sometimes if we had this mechanism. It would also be benefitial to store classes on stack in many contexts where we are not in control of everything, like when using external libraries.
What's the advantage of allocating large objects on the stack? You would get equal performance from anywhere in memory once it's in the cache. I assume cache can hold the stack + other blocks. C# does provide mechanisms to construct contiguous memory objects now. Fixed for primitives and InlineArray for custom structs. So what can't you do now?
@@gavinw77 Who said "large objects" anyway? But regarding to the cache mention, when the objects are spreaded in memory because of indirection (what usually happens with most C# classes) the chances of you having cache misses increases a lot. As of your mention of these mechanisms, I'm not aware of any mechanisms that can work with , structs where ever contiguous in C#, but are spreaded in memory and can only be accessed via indirection.
MS is adding more and more such features anywhere they find it safe to do so. The new record type, spans and memory types are examples where they provide tools for the developer to tell the compiler its safe. And you have unsafe and stackalloc for some cases, but as you say, that is unsafe ;)
I really like Rust, and C#. Seems like an ideal duo of languages to learn - one for LoB apps, the other for back-end, automation and IoT apps. Hoping MS makes Rust a first-class citizen in VS soon 🤞
They've recently hinted at that. They didn't mention VS but they did say first-class tooling. For now it's basically VS Code or the JetBrains IDEs. RustRover is their new one but it's still in eta. The VS Code implementation is pretty good but if Rust is added to VS itself, more will be tempted to play with it.
I wanted to take a look at WASM and Rust, so I worked through the excelent "Game of Life in Rust and WebAssembly" tutorial.
As they stated, the reason for Rust is not performance, but memory safety. The same reason why Google and co is adopting rust.
yes please on the Rust intro!
Will be interested to hear about "Infrastructure as code in C# dev life"
Idk, I love C. But Rust is more like C++ which I hate. Still, it removes the stupid stuff from C/C++ like the lack of a package manager or the need for header files. But the Rust syntax is still the same unaesthetical C++ garbage. IMO the borrow checker is the actual selling point. Everything else could be done with unsafe C# code as well.
Where does this put c++ in the future? Is c++ going to disappear?
C# is a great general purpose language with a very wide range of applications, but memory management is by garbage collection. If memory is critical than Rust is a better choice than C++
It's interesting they're choosing Rust. It kind of make sense from a corporate perspective since (correctly written) rust can guarantee a bunch of things about your code. However Rust also prevents you from doing a lot of things which are perfectly legitimate and it's also possible to write thing that more or less bypass the safety of the borrow checker. Hopefully senior devs won't be doing the latter, and the former is probably an ok tradeoff given their priorities.
The limitations are most likely to allow the compiler to validate the code completely. In C++, if the dev is good enough, they can do things that would look super fishy because they know it will not break. But do not let any one with less experience touch that code :D. With rust its at least harder or more obvious that your doing strange things, similar to the unsafe way in C# where you deliberately bypass the safety of the language to do things faster, but your also on your own, you can break things quite seriously that way :) In C and I guess C++ (only coded in C before switching to C#) its very easy to create unsecure code with no obvious red flags visible.
Microsoft has devs in basically every programming language there is 😂
And some you probably never heard of :)
Id love to see a Rust intro by you Nick.
Yes make one video, where you will use Rust for Web App in place of JavaScript
Rewriting C# into Rust is a ridiculous idea. I think you're far better upgrading to latest .NET version and profile/optimize exsiting code into oblivion. This is a weird signal from some areas of Microsoft. It's ironic thinking not so long ago node.js and Electron apps were all the rage. Now rushing to the other extreme. Look no further... C# is the sweet spot.
Yes to the RUST video for C# debs
I'd be very interested in a Rust intro video
The best reason to believe Microsoft is not abandoning C# is looking at all of the apps they have made that don't use Microsoft tools. With that said, often I think Microsoft's tools have adoption issue because they don't always have strong buy-in from internal teams. I believe this does have a negative affect on both outsiders adopting Microsoft's tools, and Microsoft not putting as much focus needed on the tools to make them competitive.
Yes , Rust intro video for C# devs is a Must (rhyme intended) 😛 Also a Python intro video. The above two languages are unavoidable in the long run for a well rounded C#/dotnet engineer. So at least an intro is a must.