https://youtu.be/Pzl1B7nB9KcEverything he said in this clip still holds true to this day, and everyone who attacks his character or uses ad hominems because "he's wrong", show that it'll never improve as the desktop Linux community became so narcissistic they'll never admit a fault of theirs, becoming the #1 reason why desktop Linux will never compete with Windows no matter how bad it gets.
It's not immediately apparent to me how the desktop experience is supposed to suck on Linux.KDE by the way
>>109010425Works on my machine.
>12 years agoDesktop linux is amazing now compared to then, and even if it hadn't improved much Windows is so fucking unbearably dogshit that it would have improved by association.
>>109010425As usual, Linus is 90% right, but simultaneously ignoring a pain point in the future.He was absolutely right that distros tweaking glibc fucks everything up, and that distributing binaries is a massive pain in the ass. Flatpacks sort of "solve" that these days, and the storage costs with having 3 million redundant copies of the same code don't really matter overly much anymore. In some aspects you can even argue that shared libraries are detrimental because it makes having different versions of software more complex and painful, which is another appeal of containerization.On the other hand, his obsession with never breaking user space has created some pants shittingly stupid technical debt within the linux ecosystem. Breaking user space should be minimized, yes, but blocking any changes means that you wind up with legacy shit you can never clean up or get rid of. FreeBSD has a lot of issues, but being willing to break user space means that they got fixes for various network tools sooner, or they were ported directly to the main tools rather than needing special flags. OpenBSD goes way too far in the other direction and creates headaches all the time because it'll break compatibility at the drop of a hat.The consistency guarantees made linux the top dog, but it will eventually come to a head where the tech debt outweighs the benefits. No idea what happens then. Maybe we got a bunch of kernel forks. Maybe some other ideology takes over. Maybe skynet will kill us all before then. Your guess is as good as mine.
>>109012503Dont worry, Microsoft is working round the clock to turn linux desktop into dogshit too.
>>109012804They'll never touch openbox so I'm safe
>>109013001Think again
>>109010425i'm using linux right now
>>109010425If you had actually watched the video, you'd know that the main reason GNU/Linux is not more popular is the lack of preinstalls on machines being sold commercially, and Torvalds correctly points this out.If you have a real argument other than "just watch the video", now is the time to reply.
>>109013269Cool, if there were preinstalls then a) the Linux community would shit their pants that it's Fedora or Ubuntu instead of whichever niche distro that's "superior" and b) the remaining pain points such as glibc breaking shit, shared libraries being an outdated concept that cause more problems than they solve nowadays and the absolute shitfest that is package distribution would remain and people would quickly complain about these preinstalls.It's a common fallacy to act like Linux is being held back by not being shipped on consumer devices by default, but the truth is that if it were shipped in the state that it is, even today, people would hate it for being a headache to use, not to mention lack of certain software which is the main make or break of an OS. First, fix the OS ecosystem to make it friendly to both users and developers, then you can dream about it being shipped on devices.
>>109013310It's fine as it is. Works on Chromebooks just fine. Crostini is a first class feature on ChromeOS. System76 also exists. Most applications are web-based these days anyway, even Adobe CC is. AutoCAD is the main holdback, but you can run that in a Cloud PC. It's really a non-issue.It's ready, it's finished, it has a lot of security problems and the scope of supported hardware is probably too broad at this point but there's very little left to do
>>109013318>Works on Chromebooks just fineNow you're doing the disingenuous arguing that "Linux is a kernel so Linux is any OS using the kernel" when if you had actually watched the video and understood what Torvalds was saying, he was talking about x86 desktop GNU/Linux distributions utilizing free desktop components. ChromeOS is a completely different OS that only bases itself on the kernel that has a completely different structure from the distros Torvalds was talking about, hence why none of his points apply to Chromebooks.You might as well argue that Linux has already dominated the world with Android so there's no reason to keep moaning about Windows dominance but both you and I know that's not what's being discussed here. By bringing up ChromeOS in a discussion about desktop Linux distros you're being a disingenuous liar and you know it.
>>109012705He's exaggerating when he says "never". There's plenty of instances you can read in the mailing list where he agrees to break an ancient feature and see if it affects more than one person. What he's trying to get across is that this is not something that is done willy-nilly.
>>109010425I wouldn't mind talking about this and agree with the criticisms about Linux desktop issues, I even just read an article earlier today about this:https://gardinerbryant.com/hammers-without-handles-linux-ux-sucks/but you've decided to join, "Linux as a desktop has built up embedded difficulties that have developed over time, and will be very difficult to remove" with "Linux as a desktop sucks because everyone is a bunch of evil stinky meanies that intentionally made it suck out of egoism". That's loading the conversation before it even begins and is really dishonest when you treat it as an implicit assumption I assume you think we're meant to just take at face value, given you haven't explained what this narcissism of the community is or examples of it. So I will not engage.
>>109012705>distros tweaking glibcPretty sure it's glibc breaking ABI that's the problem. GNU expects people to recompile, which works fine if all you care about is Free Software. But proprietary software that requires a stable ABI gets broken by GNU's changes. That said I am no expert, so someone that knows more can chime in.
>>109013334It is based on gentoo. your rant deserves no further response
>>109013318>AutoCAD is the main holdback, but you can run that in a Cloud PC. It's really a non-issue.>Run windows remotely so we can avoid running windows locally.>This is better because?????The absolute state of this board.Never mind that for anything military you'll literally never get all of your peripherals working correctly for drafting or 3d-modelling work because trying to get a DoD contractor to let you do that may literally require an act of congress, it's going to be unresponsive and shitty when you could just run windows locally and manage everything sanely with group policy settings. The delusion required to suggest this kind of retardation.
>>109013380It doesn't even have to be proprietary. Why shouldn't someone be able to redistribute a binary and have it Just Werk�I've got a list of issues with windows that makes CVS receipts look like a post-it note in comparison, but you know what? I can take some old piece of shit utility made in 1997 that does some stupid thing I need for work, and it'll just run. It may need compatibility mode turned on, and if it's expecting full screen or has stupid logic based around slow processors that can act strangely on modern systems, but at the end of the day it will run.That's all 99% of end users care about, and it's all effectively all management cares about. Will it work? The answer with linux is way too often no. Or if you're a FSFtard, gno. The average person doesn't give two shits about free software, and nothing you or anyone here says or does will change that.
>>109012705>On the other hand, his obsession with never breaking user space has created some pants shittingly stupid technical debt within the linux ecosystem. Breaking user space should be minimized, yes, but blocking any changes means that you wind up with legacy shit you can never clean up or get rid of.No, Linus is right. Linux should be more like Windows in that you should stop creating unnecessary work for application developers and make sure that everything that used to work continues to work.Unfortunately for the Linux OS as a whole, the developers of userspace libraries completely shat all over Linus's principles and that's why Linux sucks and is barely usable for desktop computing
>>109012804They literally don't have the power to do anything.
>>109013380The biggest glibc abi breakage concern a feature that had been deprecated for 10+ years before being removed. But yeah, it's mostly proprietary software (EAC in that case) that complained because their entire workflow relied on that particular deprecated feature.https://bbs.archlinux.org/viewtopic.php?id=294184For 99.9% of usecases, glibc is forward compatible, that means if your program was compiled for glibc 2.2, then it will run on glibc 2.3 systems.However, the optimal solution is always to let distro maintainers build and publish your program. There's an entire class of programs that can't be containerized. Everything that touches OEM-specific graphics/computing api, like cuda, shouldn't be containerized because you have to install drivers and computing libs inside every containers.If you're building a niche program that won't make it in distro repositories, the solution is either: - let users build it - statically link everythingThe second option is almost always the worst, except if your program can be used with musl. Some libraries were built for the purpose of being dynamically linked (openssl for instance) in order to get security patches without having to recompile your entire application and redistribute it when the library gets updated.There are many other reasons why this could break, for instance graphics API need to talk to your GPU's ICD which can't be statically linked as there may only be one ICD at once.Languages like rust have chosen the first option, because they have their own source-based package manager to install and build programs and dependencies.You also benefit from architecture specific optimizations this way, though it's not as significant as it used to be 10-15 years ago.
>>109013417Sometimes I wonder if /g/ userbase is made of primarily mentally ill children.
>>109013738Given how the average quality of any tech discussions, especially talking about OS architectures, computing logistics and other complex autism, is on par with your usual YouTube comment section infested by Clippy avatar midwits, I'd say you're onto something.
>>109013738I'm NOT a child.
>>109013936Yes you are, child. Enjoy prison.
>>109013946Mentally? Maybe. Emotionally? Perhaps. But legally speaking, I am a man.
>>109013496Oh you sweet sweet summer child.
>>109010425Who cares about what Linus says about software, silly techtuber can't even install popos
>>109014376funni, but I explicitly referred to him by his surname to avoid this ol' schtick.
>>109014358You are retarded if you think they do.
It's not simpler on windows, apps have to ship with their own copy of DLLs or packages too. How many times steam is installing DirectX or msvc c++ runtime again and again to make sure the user has the needed DLLs.
>>109010425>Everything he said in this clip still holds true to this day,the one specific thing he mentioned (bumping glibc) is ironically the one thing that shouldn't break anything. because glibc doesn't bump so versions and use versioned symbols exactly because of that. so running older binaries loading a newer glibc shouldn't be a problem. the opposite is not the case of course.where glibc is actually shit, is when it comes to not playing nicely with static linking. but here is where musl (and the kernel's promise to never break user-space) come to the rescue.in CURRENT YEAR, this all can be made trivial by using Rust with the readily available musl target. with other languages, it can be done, but the difficulty level? well, it depends (dependencies, build system, ...).>and everyone who attacks his character or uses ad hominems because "he's wrong", show that it'll never improve....i have no idea who you're referring to, and i don't care. the peanut gallery is irrelevant.>becoming the #1 reason why desktop Linux will never compete with Windows no matter how bad it gets.who told you there is a competition going on?and competing over what?financials? my distro is neither selling copies NOR data (wink wink).plurality? why would your tard /g/eet self think "we" would even want that?desktop linux has been fully serviceable for at least a decade (around the period OpenGL versions started rolling).desktop linux reached a critical mass of users probably 2 decades ago.even tards are running whatever they want nowadays using container crap (since they can't into understanding real building+packaging).
>>109010494>krashes
>>109010425>What is AppImage>What is FlatpakYes he was, but he WAS right, he isn't anymore.
>>109016046>What is AppImage>What is FlatpakFurther fragmentation in already hopelessly fragmented landscape.
>>109017565And before someone jumps in to say Windows is fragmented, the fragmentation is different. You have a few different convoluted ways to have basically the same end installation, be it in Program Files or other folder, with any necessary registry entries being added.In Linux, practically every form of software distribution results in a different end installation. Package managers do the traditional installation, AppImage aren't installed but simply ran, the whole program is contained within a dedicated file. Flatpaks are installed in it's own little containers.That's three separate forms of software installations on one systems. On Windows you always have one form of installation, even with UWP's. There's a folder with all the necessary files and the main executable that you run. No containerization, no portable packaging, just loose files in a location. Basically what you get on Linux with traditional package managers, but those do it worse and every attempt at fixing it only made things more complicated than they should be.
>>109013006>flatpakNgmi
>>109015060Literal microsoft employees are merging code into systemd, used by every major distro.
>>109010425>Everything he said in this clip still holds true to this dayNo, it's not. We now have Flatpak, AppImages, containers etc, statically linked shit like Rust and Go. It's trivial to make an executable that will run on most distributions.
>>109017812>No, it's not>proceeds to miss the point proving him rightOnly the brightest minds of /g/
>>109017818I guess you didn't watch his video, retard. His point was that developers had to make hundreds of executables for each distro. Now they can make only one set of executables and ship them to most distributions. Eat shit, faggot.
>>109017828Suck my dick you gay fag I'll fuck you in the ass you stupid son of a bitch. Do you understand me? DO YOU FUCKING UNDERSTAND ME?? GET OFF MY FUCKING TECHNOLOGY BOARD BEFORE I FIND OUT WHERE YOU LIVE AND WRECK YOUR ASSHOLE WORSE THAN A HARD DROVE THATS NOT BEEN DEFRAGGED IN 20 YEARS FUCKER. DONT FUCK WITH ME
