Martin was out of line.
Hellwig also was out of line and unnecessarily hostile.
Linus... Is the voice of reason? Though I would have preferred he rebuke Hellwig in the same breath.
It's a strange 2025.
this post was submitted on 08 Feb 2025
192 points (98.5% liked)
Linux
5951 readers
495 users here now
A community for everything relating to the GNU/Linux operating system
Also check out:
Original icon base courtesy of [email protected] and The GIMP
founded 2 years ago
MODERATORS
Martin has nearly always been like this. Ive known martin from way way way back when he worked on the wii and he has always been a guy that just causes drama by pointing and saying "this is shit. Look at this shit". It isnt a bad thing to do, but the way he does it is basically going to somebody's home with a sledgehammer and smashing a wall without checking in. It turns people away from you even if youre right.
He had a beef and drama with me, devkitpro, gbatemp.
Then he stopped being on my radar, heard he was working on asahi, then heard he was causing drama between emulation devs, then luois rossman, and now bloody linus torvalds?
sigh
pointing and saying “this is shit. Look at this shit”
Yeah you only get to do that if you're Linus 😄
You only get to do that if you’re the founder and leader of a project. If you join other people’s projects then you play by their rules, not your own.
Linus stepped away from the spotlight and apologized and went to therapy and changed his behavior
He definitely improved it... but there's still plenty of material for r/linusrants.
He may not directly call people names anymore but he's still extremely rude and unprofessional. He would have been fired long ago from any company I work for, and I live in the UK where it's practically impossible to get fired.
I live in the UK where it's practically impossible to get fired.
In the UK you only get employment protections after working at a company for two years. Up until then you can be fired without reason
That's not true. See https://barcankirby.co.uk/two-year-rule-for-employment-rights/
There are dozens of circumstances where an employee can claim automatic unfair dismissal, or where a worker thinks they have been unfairly dismissed or mistreated due to their protected characteristics which could lead to an additional unlawful discrimination claim.
The general fear of litigation means that in practice it is really hard to get rid of underperforming employees in most white collar jobs (once they have passed probation which is normally 3 months). The normal way it happens is they are "managed out", which is a long and unpleasant process.
It is very much true
From the link you provided:
This may come as a surprise to some: it is only after two years of continued service that employees have the right to request written reasons for their dismissal.
Employers don't need to provide a reason for your dismissal.
There are some exceptions but they're very specific and unlikely to relevant. If an employer wants rid of you they can do so at their whim.
Even if you think you have a case, good luck finding a solicitor who will take on your case. They simply won't be interested because it's not easy to win without the rights you get after 2 years employment.
I'm not sure you read my comment at all...
I did.
I also read a fair chunk of the article you linked and it didn't change my view.
I disagree that employers fear litigation. Is there something else I missed?
Amazingly restrained on Linus' part. He has really grown.
Well, social media around him and us really hasn't. Torvalds is right to condemn the agitation, but Martin's case is just by-catch to the underlying technopolitical issue.
This makes me sad because I assume Hector is deeply demoralized. I've seen him as a hero since he started the Asahi project. I know nothing about him other than working on Asahi, Asahi worked when I installed it last year, and this blow-up last week or so and now. Maybe he's a terrible person. I just see his technical work (and all the other devs contributing to the project) and I know this would make me lose steam if I were in his shoes.
He didn't even say anything wrong. In all that discussion and whole thread linus could've said anything useful for either side or to help things move forward but chose to do nothing besides saying 2 things and they didn't address any of the concerns. He just contributed to the problem. One could've said that he didn't read or something, but no, he came, read everything, all the technical concerns, all the needs from the stockholders involved and decided not to address the elephant in the room!
I would've done much worse than hector and call half of the people there names.
Really people should go to the kernel threads and read all of them carefully and the links to the social media and read them careful too, it's not too big.
Anyway what matters is apple/arm work will continue out of the kernel tree, downstream.
And this could basically also lead the whole rust effort go the same direction by all the betweens lines we can understand in the whole discussion. Basically there are enough people that feel that no other language other than C should ever be in the kernel. Doesn't matter if it's rust or not. No other. Even if some people are supportive, there are enough in key positions that will fight to not include it.
Really people should go to the kernel threads and read all of them carefully and the links to the social media and read them careful too, it's not too big.
No, people who aren't involved with the kernel should not get involved with this, that's Torvald's point. You can't let a serious project like Linux be affected by social media uproar, let the developers sort it out amongst themselves.
Linus shouldn't have to get involved at all. Each part of the Kernel should be handled independently by the maintainers. Linus responding publicly to outside forces is fine but once he has to step in to handle public fights between individuals who are supposed to work together it is a problem.
Linux staying C focused is a valid thing to do. It is very hard to get folks to contribute to the kernel and if you cut out anyone who doesn't know Rust, a language with at best 5% the adoption rate of C, you will run into spots where sections of the kernel are unmaintained due to no willing and qualified person covering it.
Adding Rust based functionality and support is great. Changing APIs to require maintainers to learn Rust to continue to maintain the code they are experts in is unacceptable.
And clearly you didn't read or have been following the r4l approach, because nobody is saying to change c apis. Go read.
Claburn's article seems biased toward Martin's position in the disagreement, using the most forgiving language possible for his behavior while describing the opposing side with obviously critical language and insufficiently covering the reasons for it. Linus's response might be mildly interesting, but the article is disappointingly poor journalism.
Oh I expect absolute crickets from all the people who waded into this trashing Hellwig.
I expect you to have zero kernel commits to your name.
I will show you mine if you show me yours.
Not that number of commits has anything to do with calling out brigading which you still seem to be committed to.
Well, here is a very funny one-off commit, but my biggest effort was probably substantial parts of a couple AMD/ATI GPU drivers, well-summarized here. As usual, that was a team effort, with particular credit to Deucher (AMD), Glisse (radeon maintainer), and Airlie (DRM/DRI maintainer). So, put up or shut up. Or, to paraphrase the sentiment that you seem to not grok: talk is cheap; show us your code.
Let me make it clear. I call out brigading because it is useless noise that distorts and obfuscates the kernel development process. I don't care that you're salty that I'm pointing out that your "absolute crickets" comment is not only incorrect, but empty in the sense that your lack of perception is not a substitute for the actual process of kernel development. Additionally, in this case, it seems like you're still focused on personalities rather than the underlying computer science; I expect "absolute crickets" when asking you about the topic of memory safety.
My commits have my real name, because I do serious work, and no I won't be doxxing myself on this account.
I call out brigading
What kind of lazy gaslighting is this? You were part of the brigade. You have literally been participating in the kind of behaviour that Linus has correctly called out as unhelpful.
Please continue to think you're on the right side of this discussion when the guy you went out to bat for has stepped down as a maintainer and deleted his socials over this.
What's with the quotes?
No amount of sabotage from old entrenched maintainers is going to stop the world from moving forward towards memory-safe languages.
Jesus fucking insufferable Christ... Saying shit like this, given C has been in use for 50 years and is still in very wide use today, and given the vast number of languages that have come and gone over this period, it's just incredible.
I do not agree with the Dev who stepped down.
But on the topic of C, I wouldn't measure the quality of a language based on its adoption. C is a relatively old language and therefore benefits from getting wide-use before other languages were born. It will never die because who would ever want to rewrite every project in existence in another language.
Memory safety is very important since it has consistently been one of the largest sources of vulnerabilities throughout software history.
C is not a bad language, but it has flaws. Performance at the cost of safety is not a good trade-off in most scenarios. There is no such thing as a "perfect programmer" who won't make mistakes.
I don't disagree with these points in general. However this isn't simply about the tools. Tools go along with people and their skill and experience. There are developers and developers. There are people with lots of experience who create much higher quality C code than others. Personally I'd never touch C if I can avoid it as I don't trust myself as much. I'd always go for C++ instead. Modern C++ with RAII is great. It's what most of the software at our corpo is written in. Maybe Rust would end up becoming the default standard at some point. Maybe something else would. I would never go shit on a coworker who has produced tons of well functioning code that they better reskill in something that may or may not stick around, or that they may not become as productive with for a long time. A team skilled in C or C++ may be able to produce higher quality software, quicker than a less skilled team Rust. Rust might be better for teams that just start in native programming. I don't know. If it grows enough in use, reskilling people and reworking software to cooperate with it might become an obvious choice. For now, as I see it, it depends on the team.
And I dont deny that. There are a lot of programmers, and not all had eduction on designing secure software. Even with the knowledge and experience, what if the programmer is tired or makes a similar mistake. Only one mess-up away from a potential vulnerability or instability of the app and system as a whole. I need more experience with C to form a better opinion.
This is why security is usually multi-layered - decrease the chances of a single fuckup compromising everything. And yes using a safer language adds a layer. But typically it won't be the only layer.
Memory safety at the expense of complicated interfaces is also not a good trade-off, even in terms of security.
Probably what all the horse people said when cars were invented.
I would't trade a good horse for an early car. Maybe a model T.
The horse-car analogies rarely achieve what you want them to, especially in situations where we don't have the benefit of hindsight.
More akin to Elon insisting the cybertruck is the way of the future, and people just keep buying Rivians and internal combustion vehicles.
Especially as there's D, a language that when used in betterC mode, is on-par with C, minus the archaic precompiler, which is replaced with very powerful metaprogramming capabilities.
D didn't catch on for many good reasons... And it never will. Zig has way more momentum (and it better!) as a "better C", and obviously the main draw of Rust is memory safety without GC which D doesn't have.
D has a way for GC free operation, the easiest way is by using it in BetterC mode. The harder way is writing your own runtime without a GC.
Yes but that isn't memory safe.