Linus Torvalds Discusses Maintainers, AI, and Rust in the Kernel (zdnet.com) 31
ZDNet reports that "At the Linux Foundation's Open Source Summit Japan, Linus Torvalds and his good friend Dirk Hohndel, the head of Verizon open source, talked about the current state of Linux:
Speaking of maintainers, Hohndel brought up the question of "maintainer fatigue and how draining and stressful this role is...." Torvalds replied, "It's much easier to find developers; we have a lot of developers. Some people think that you have to be a superdeveloper who can do everything to be a maintainer, but that's not actually true...."
Hohndel commented that the aging of the kernel community is a "double-edged sword." Torvalds agreed, but he noted that "one of the things I liked about the Rust side of the kernel, was that there was one maintainer who was clearly much younger than most of the maintainers. We can clearly see that certain areas in the kernel bring in more young people...."
Hohndel and Torvalds also talked about the use of the Rust language in the Linux kernel. Torvalds said, "It's been growing, but we don't have any part of the kernel that really depends on Rust yet. To me, Rust was one of those things that made technical sense, but to me personally, even more important was that we need to not stagnate as a kernel and as developers." That said, Torvalds continued, "Rust has not really shown itself as the next great big thing. But I think during next year, we'll actually be starting to integrate drivers and some even major subsystems that are starting to use it actively. So it's one of those things that is going to take years before it's a big part of the kernel. But it's certainly shaping up to be one of those."
Torvalds also said he enjoyed the fact that open source "has become the standard within the industry."
But later Hohndel, calling AI "autocorrect on steroids," asked Torvalds if he thought he'd ever see submissions of LLM-written code. "I'm convinced it's gonna happen. And it may well be happening already, maybe on a smaller scale where people use it more to help write code." But, unlike many people, Torvalds isn't too worried about AI. "It's clearly something where automation has always helped people write code. This is not anything new at all...."
But, "What about hallucinations?," asked Hohndel. Torvalds, who will never stop being a little snarky, said, "I see the bugs that happen without AI every day. So that's why I'm not so worried. I think we're doing just fine at making mistakes on our own."
Hohndel commented that the aging of the kernel community is a "double-edged sword." Torvalds agreed, but he noted that "one of the things I liked about the Rust side of the kernel, was that there was one maintainer who was clearly much younger than most of the maintainers. We can clearly see that certain areas in the kernel bring in more young people...."
Hohndel and Torvalds also talked about the use of the Rust language in the Linux kernel. Torvalds said, "It's been growing, but we don't have any part of the kernel that really depends on Rust yet. To me, Rust was one of those things that made technical sense, but to me personally, even more important was that we need to not stagnate as a kernel and as developers." That said, Torvalds continued, "Rust has not really shown itself as the next great big thing. But I think during next year, we'll actually be starting to integrate drivers and some even major subsystems that are starting to use it actively. So it's one of those things that is going to take years before it's a big part of the kernel. But it's certainly shaping up to be one of those."
Torvalds also said he enjoyed the fact that open source "has become the standard within the industry."
But later Hohndel, calling AI "autocorrect on steroids," asked Torvalds if he thought he'd ever see submissions of LLM-written code. "I'm convinced it's gonna happen. And it may well be happening already, maybe on a smaller scale where people use it more to help write code." But, unlike many people, Torvalds isn't too worried about AI. "It's clearly something where automation has always helped people write code. This is not anything new at all...."
But, "What about hallucinations?," asked Hohndel. Torvalds, who will never stop being a little snarky, said, "I see the bugs that happen without AI every day. So that's why I'm not so worried. I think we're doing just fine at making mistakes on our own."
Re: (Score:2)
You don't throw away stuff that works, especially when it's "well known" how it works.
Replacing it with something new that is unknown is risky. We have a "new guy" on the team, this guy keeps wanting to use all these friggen cloud tools that cost money instead of ... oh I don't know, using a text editor and spreadsheet we already paid for.
And connecting all the API's of all these cloud tools? Also money. Why? I can write all this stuff myself? But to avoid re-inventing the wheel I'm like "fine, you do you,
Re: Rust is a bad thing (Score:2)
Until you publish your own port of the Linux kernel written entirely in Fortran, I'm going to say that you don't even take your own advice.
Re: (Score:2)
"Not invented here" tends to be what drives a lot of software development.
Instead of just buying the wheel, someone goes "I want it for free" and just xeroxes the dimensions of the wheel and then goes out and re-invents processes for making rubber and steel belts, and makes their own hand-made tire molds, one tread at a time. Y'know, instead of just buying the wheel. More power to you if you want to do this for funs and giggles, but you're going to find it seriously annoying when you have to do this 4 times
Re: (Score:3)
The cool thing about open source is if that one guy stops working on the project, literally anyone else can pickup where he left off. If this guy poisons the code and leaves the project, we can always fork the last version of it and go from there. All do able with open source.
On the other hand, if a company, oh let's say Alphabet, writes an important application (LOL but let's pretend) and gets a lot of following, then cancels the project, you can't even fork it from the last release because it was propriet
Re: (Score:2)
Re: (Score:3)
I find it's a shame that most young people seem to be interested only in "new" stuff.
A failing of the young: No respect for things that are proven and work. Some are smarter, but most young people must make the same tired old mistakes over and over again to realize things they could have learned directly. Some never learn and forever keep chasing the silver bullet technology that will finally make their skills not suck. Incidentally, proper engineering education is not like that. It is just the excessive immaturity of the field of software engineering that leads to this stupidity.
Concerned about AI-generated code? (Score:4, Insightful)
Why would Torvolds be concerned about AI-generated code in the kernel? Maintainers scrutinize code submissions, that's what makes the job hard. It doesn't matter where good code, or bad code, comes from, what matters is the code is judged on its merit.
Re:Concerned about AI-generated code? (Score:4, Interesting)
It'll get really interesting when it's AI that's doing the judging.
And it will happen, to some degree, eventually.
Re: (Score:2)
Re: (Score:2)
Funny thing: For statistical models (what drives the current hype), hallucinations cannot and will not go away.
Re:Concerned about AI-generated code? (Score:4, Interesting)
Re: (Score:2)
AI generated code could, and will likely have, much more subtle defects and vulnerabilities than regular code. That comes from the deceptive nature of the currently hyped type of AI: It is quite good at projecting a sense of competence while being completely clueless and a sense of correctness while generating excellently looking bullshit.
With a regular crap developper (the usual kind), you can immediately tell they are not good at it because they make basic mistakes. With AI, those clues fall away.
Re: (Score:2)
Human-generated code can have subtle defects and vulnerabilities just through accidents.
If there is a risk of AI-generated code being more vulnerable, it is only because the AI code generator is not intelligent. It might mix code samples which were given to demonstrate a vulnerability, or even code from an obfuscated programming contest, if it somehow got into the training data.
On the other hand, a human might do the same thing and not even know it. "I saw this code somewhere, but can't remember where"...
Yes the Cat has my tongue... (Score:2)
He wouldn't rust in the kernel if he kept it properly lubricated... Just sayin'
Re: (Score:2)
Which is why the programmers toolkit includes WD40 along with the duct tape.
AI In the Kernel. (Score:2)
Rust in the kernel (Score:2)
If Rust goes in the kernel, how will it handle process suspension, because, as we all know, Rust Never Sleeps
Re: (Score:2)
Well, I for one, plan to drop Linux if Rust should ever become mandatory in the Kernel. These people are hallucinating about what their shiny new tool can do.
Re: (Score:2)
Re: (Score:3)
* "Microsoft is rewriting parts of the Windows Kernel in Rust" https://www.thurrott.com/windo... [thurrott.com] "Microsoft continues its Rust mission with new kernel features" https://www.itpro.com/software... [itpro.com]
* "Apple is starting to use Rust for low-level programming" https://news.ycombinator.com/i... [ycombinator.com] Not clear if they include Rust in the kernel, opinions are that it's possible to be already there due to the sheer size of the compa,y and different options of the teams.
* Resources on how to write FreeBSD kernel modules and
Re: (Score:2)
I'm curious what your opinion would be after you've actually used Rust.
Rust is a bad idea? (Score:2)
The potential problem with Rust in the kernel, imho, is that the number of potential Rust devs is significantly smaller that that of C or C++ devs (according to the internet, Rust = 0.5 * C++). A related problem is man-years of experience.
Might not matter if you have $inftinity to hire devs, but if they have to work for free...