Beginner friendly deep dive into the latest kernel! Thank you for watching! Part 2 0:00 Intro 0:48 …
[ad_2]
source
Beginner friendly deep dive into the latest kernel! Thank you for watching! Part 2 0:00 Intro 0:48 …
[ad_2]
source
Lorem ipsum dolor sit amet, consectetur.
Allow me to state my opinion more clearly.
Rust can be useful in the kernel.
If you want more Rust in the kernel, you should try to get more maintainers to know Rust, and the best way to do that is to get most NEW maintainers to know the language.
You can also try to get existing maintainers to learn the language, but this is unlikely. (PLS don't go and think this statement is malicious, if a maintainer says no, it's no).
The way I explained it in my video wasn't clear, and it sounds ageist, and it's generally inaccurate, age doesn't change whether you know Rust or not (;
Now you can all debate whether Rust should exist at all XD
More and more I feel Zig would be a better choice for a new language in the Kernel.
Thanks for your Videos :3
BSD
I don't like rust, it's ugly and verbose. Next to C it looks like drugs. Unmaintainable. I'll be disliking the video advocating for it.
Rust is needed
Very nice commentary format to get a quick dump of what's happening in the kernel!
… why not make another OS completely in RUST?
seems like you are doing some Frankenstein stuff with all these combinations…. not really elegant.
Writing a kernel from scratch would look like a much sounder approach, there's already one in Rust, why don't they focus on this instead of trying to "invade" Linux (I know why: money)?
As to languages, Rust is certainly not what I would have picked for a new kernel, if I wanted to try something else than C.
Why do you pronounce Linus' name like [lainus]? It's [li:nus], just like [li:nux], (as in leenus/leenux).
I think zig would be better than rust imo
You want old programmers to step aside, but there will be tears when they go away…
No solution to rust drama?
Write code in C.
Write documetation.
Ask rust devs for help to fix broken part using documentation.
Asking for help is OK!
18:00 you don't need to know Rust to write proper documentation and commit message to your C code.
You are doing a great job with this summary. It's very instructive, nicely explained, not boring. This must be quite a lot of work. Very nice. Thank you!
Maybe this is the wrong place to ask but I'm asking anyway. lol For those who like having Rust in the Linux kernel, how well do you think the Rust Linux kernel developers know 'C'? If they know Rust much better than C, I can understand hesitation by the C developers to incorporating more Rust into the kernel. If the Rust Linux developer is as proficient in C as they are in Rust, I think they should be able to make a more convincing argument to adopt Rust since they can advocate in 'C-like' terms/perspective vs using "because it's safer" kinds of arguments.
I wonder, when Linus passes away if there will be more big changes to the kernel. I think Greg Hartman will add native support for a standardized configuration system across distributions to specify automount behavior, replacing /etc/fstab. He's fucking crazy man!
I want a filesystem with native encryption except zfs for the love of God!
Rust in the kernel started with a weird twitter campaign foreshadowing personality clashes. I think it was Linus' first big fumble. I also disagree with the mentality about programming that Rust represents which make people think it's superior.
So the people finally got fed up with the influencer language bs. Good riddance.
Rust and C devs blaming the other side, need to look in the mirror. They are both the problem. Tribalism is real. I say this as a C dev.
Changes shouldn't be pushed faster than the natural rate of your senior incumbents naturally adopting it. If you want to, let it develop independently of them. Every system of every type crashed hard, often destroying everything good that was already there with no path to become good again but within a false perception bubble, forcing change faster than natural adoption and adaptation.
The drama is about Old fart that don't want to go back and documents their spaghetti code
I don't get why anything unstable or projected to not be maintained in a continuous stable working state for at least several years ahead would be in the mainline. Let people who want it inline merge it into their custom kernel.
Let's use only Rust in Security and Crypto. Let's use only new, fresh maintainers for it. I understand there is a great training program for them in Quantico.
Ntfs using the ntfs3 driver from paragon that was merged into the kernel ALWAYS makes errors in the filesystem that have to be fixed from windows while the fuse ntfs-3g extremely rarely does that. Literally ntfs3 mounted drive i have to check consistency every week and every time windows finds problems. With ntfs-3g it's super rare. I wonder if anyone else experienced this. Both on latest kernel 6.10.x or the 6.6 LTS kernel it doesn't make a difference
Huge fan of your takes in this vid. Super agree.
What I don't like about Rust is that in contrast to C, which is very simple language, standardized by ISO, has at least 2 big compilers (and really lots of small), Rust is young, quite complex, has huge corporation behind, and only has 1 mature compiler. It more similar to C++ than C, which is hated by Torvalds and not in the kernel. I totally understand why people like Rust, but I wouldn't write significant software that is supposed to exist in long term in this language.
tldr: In my opinion, for (probably) the most important software in the world, Rust looks as big and unstable dependency
u should start yr podcast , really good content btw
When the โold developersโ retire, one has to hope that Rust is still alive. Religions come and go.
I think party of the rust problem is a cultural one. In Rust, you put as much as possible about the semantics into the type system, so that the compiler is responsible for enforcing the semantics. Some traditional c developers have the approach that it is primarily the developer's responsibility to know how to do and use things. The rust facades for kernel c apis trigger everybody because that's where the cultures meet.
real programmers can write multiple languages. people complaining about it is hella cope…. git gud.
promote one tool over another while almost does the same thing is just, politics
always rust believers promulgate that rust is superior, and without question, it's just a law for you people.
i, however, bag to differ, and i don't have a problem presenting my claims.
As a C developer, even before rust, i could always switch to C++, but i didn't because i like the fact C is minimalist and i also claim it's objectively safer and potentially (also depend on the dev skills) more performant than C++ (or rust, for that matter) – due to forcing the developer to write code on his own, tailor made to his program's needs, instead of using pre-made general Swiss army knife function.
The rule of thumb here is the more functions and keywords a language have – the higher the attack/fault surface as well.
This is quite evident in the changes both C++ and rust went through, throughout the time they exist.
Obvious examples for the issues with kitchen sink languages are both php and python – take for instance the breaking move from php 5 to 7, or python 2 to 3.
rust will be no different, with all the functions it supports – some will always break in the future – that's the price a developer will always pay for taking the shortcut of using someone else's ready-made code.
beyond that, my experience with rust is that it's not even close to C in terms of memory usage, not for the RAM and not even the project's source code has i had small projects in C reaching to 3-4 megabytes, while rust, with all the cargo build i ran, bloated to a 4GB of disk space, and after cargo clean was still over 400 megabytes.
For RAM usage, i experienced 2-4 times larger usage than the same C project.
Last point i'm going to mention this time regarding rust, is that it's not even safe as people think it is.
When i wrote in rust, i started noticing that every small project is pulling dozens and up to hundreds of dependency packages, and i'm talking about really small projects.
The problem is, when you add 1 small dependency, it's highly likely that dependency will pool it's own dependencies which in best case scenarios i've encountered, accumulated up to a dozen more deps.
This is a nightmare from security POV, as you need to audit each of these packages, not to mention doing so upon each and every upgrade of them.
The same rule of thumb applies here as well – the more code, the higher the attack/fault surface as well.
There are other points to make, but that's already to long
we don't need rust in the kernel
Where's your old videos?
Heck Rust. Go write your own kernel, stop Star Warsing the best open source project there is.
You said "old developers should quit" because you want more Rust in the kernel… I'm using Rust, but I'm trying to avoid any contact with a community because of statements like this one. Please refrain yourself from a private opinion that someone might interpret as an insult ("Are you an old kernel developer who uses C and doesn't like the idea of implementing more stuff in Rust? then just quit."). I would be thankful if Rust developers working on the kernel wouldn't think about Rust as some type of cult.
Rust is the future of the kernel, that's clear. C would work perfectly if everyone was a perfect developer, but they aren't and we're seeing the quality of developers drop with younger generations, especially Gen Z that aren't as technically competent as Gen X and Millennials. Rust essentially provides the safety wheels by being more verbose and having an aggressive compiler that checks everything.
It's necessary to make Rust the core of the kernel because we simply are losing all the good developers to age and we can't expect the new young developers to be as good so we have to rely on Rust to essentially force them to write good code (against their will).
What happened to NTSYNC?
This is a great review of the kernel changes. Love this style way more than the ones I've seen on bigger channels. Subscribed for more kernel update summaries ๐
I really appreciate these filesystem update sections!!
Love learning about filesystems
I don't agree with the point of having the old devs quit. I agree that having 2 languages for the kernel is annoing for some people but it should not mater in the long run. The kernel used to have, and still does I think, a lot of assembly. it was a hassle to mantain but people did it. I think we see a lot of growing pain and thinks will smooth out in a few years as the older dev get used to rust and the newer ones get used to C.
Finally 6.11 is released, i can now use my laptop's keyboard without using an rc kernel!