Gregory's Server@lina @desttinghim what bothers me about this is it also feels like a leadership failure. The reality is if someone like this doesn't see "the value" in rust then they should be required to find someone else who can be trusted as a maintainer to handle the code. I don't know how folks like Linus actually expect rust to pick up in the kernel if we don't address the maintainers who clearly want nothing to do with this. There's plenty of other contributors capable of reviewing this code
|
24 comments
@Lyude @lina @desttinghim yeah, if it's prompting pushback so intense people are explicitly giving up on "the future of the project" the people causing the pushback need held accountable  @Lyude @lina @desttinghim If we take Rust out of the equation and look at any other large scale modernization effort, we see pretty much the same problems. I like to pick on folios because I'm mostly a storage guy, but that transition has been long and painful because there's fundamentally no real leadership in Linux. A lot of storage developers just cannot handle working in an organized fashion. DRM is the most organized subsystem and even it suffers through these kinds of problems. @Conan_Kudo @Lyude @lina @desttinghim I believe that anything that is community driven and involves some many people ends up have some sort of sluggishness in the process. For me, democracy ends up adding these long and painful transitions. We can see that in our Congress, where we do have leadership, but we don't have a dictator. From what I see in the kernel, we do have leadership, but we don't have a "benevolent dictator" that points us to a single direction. @Conan_Kudo @Lyude @lina @desttinghim One might say that other open-source projects don't have this issue. But we need to remember the size of the Linux Kernel project. I can't think about a open-source project that is as big as the kernel and has such a diversity of people (from all ages, genders, colors and cultures). It is different when we have 150 active contributors in a open-source project and when we have more than a thousand. @mairacanal @Lyude @lina @desttinghim Leadership comes in many forms, it doesn't imply a BDFL. Though note that Linux *does* have one in the form of Linus Torvalds. He just doesn't use that power much. But okay, leadership in a comparably sized project? Kubernetes qualifies. It might be overkill for many infrastructure architectures, but it definitely has *thousands* of contributors, an organized leadership for every part through SIGs with good feedback and decisionmaking processes. It can work. @mairacanal @Lyude @lina @desttinghim Maybe if there was some kind of even semi-formal decision-making and community engagement process for the kernel. Some kernel developers sometimes don't even like talking to each other, much less the people that try to use and contribute to it. Most kernel subsystems fail the test for a structured subcommunity, so no I don't think they qualify. @Conan_Kudo @Lyude @lina @desttinghim I feel that saying that "Some kernel developers sometimes don't even like talking to each other, much less the people that try to use and contribute to it" isn't really helpful to create a nice relationship with kernel developers. I also must point that this is your impression about kernel developers. Correct me if I'm wrong, but you didn't do a research between developers asking them those questions. @Conan_Kudo @Lyude @lina @desttinghim Usually, kernel developers (especially maintainers) are just trying to do their best in their limited time. @mairacanal @Lyude @lina @desttinghim I spend a lot of time interacting with kernel developers, so I *do* know. Mostly I'm in storage stuff, but I do pop up in other places too. I even have started going to Linux kernel conferences and engaging there. I was at the event that video referenced in the email that started this whole thing. @Lyude @mairacanal @lina @desttinghim Frankly, I'm a bit insulted that the assumption was that I don't know anything about the Linux kernel community. I've been a known quantity in that area for almost a decade, and I've been heavily engaged in several areas over the last couple of years thanks to the Asahi Linux project. What's frustrating is that this is a problem even when I engage with kernel developers who *already know* I have some know-how, but don't care and ignore my feedback anyway. @Conan_Kudo @mairacanal @lina @desttinghim I get that. honestly - it's wild how just about every time I talk about a subject like this people assume the same thing. It's incredibly rare for me to see someone disagree with statements I make like "people should behave" and not follow it up with some sort of assertion I'm new to this and simply Do Not Understand Complicated Kernel Development. @Conan_Kudo @Lyude @lina @desttinghim I don't think anyone had the intention to say that you don't have any knowledge about the kernel community. At least, on my side, I was just trying to say that we shouldn't generalize our personal experience and also, we usually have a hard time building constructive feedback when we say that the other part doesn't want to hear. We need to improve our community, but just pointing everything we do wrong makes it harder to open space for positive outcomes.  @Lyude @lina @desttinghim I kind of hope Linus himself will master Rust and introduce a bunch of it in quite critical points. @raulinbonn @lina @desttinghim It even extends outside of just being obstinate - subsystems like input have become pretty difficult to get a lot of work into simply because there's not enough maintainers, yet despite this being a problem for ages now I've seen basically no effort from the single maintainer there to actually bring in new maintainers when other subsystems like DRM very rarely ever have only single maintainer for anything @Lyude @raulinbonn @lina @desttinghim It's a big part of why @AsahiLinux has made no progress on upstreaming in over a year. The people who worked on it have become really demoralized due to lack of support from... well, anybody in the kernel community, even with @torvalds using it for a brief time. (I don't even know if he still does or if he cares about it anymore...) @Conan_Kudo @Lyude @lina @desttinghim @AsahiLinux @torvalds So particularly frustrating because the DRM subsystem had room for improvement given the bugs Lina was looking into. Estability/quality improvements there would benefit way more than just Asahi Linux and the Mac devices @Lyude @lina @desttinghim About @torvalds himself possibly coding in Rust, well, slightly fat chance. Linus just months ago explained in a talk that he is not much of a programmer these days. His day to day "technical lead" job is rather merging other people's code, aka managing code, but not really writing it: @Lyude Huh, how'd I end up in the mentions here? It's not really a problem, I just think it's weird. Is it because I boosted it? |
@lina @desttinghim yet: the sentiment we seem to agree on is that adopting rust is quite literally important to the future of Linux, since it's not just about the benefits the language brings: but the fact it actually makes kernel development appealing to a wider audience and ensures it remains sustainable