1. 10
  1.  

  2. 44

    There is very little in this I can agree with, except the last part.


    Re: “Leadership under attack”:

    Neither RMS nor ESR have never had any significant involvement in Linux, so why bring these up in the context of the Linux kernel? Seems odd.

    Anyway:

    • RMS has been highly controversial for a long time, as he mentions himself. He probably turned off more people from Free Software than attracted. He was always a highly problematic figure.

    • Eric S. Raymond hasn’t been meaningfully involved in the OSI for a very long time, and when he came back it was for little more than for a weirdly misplaced rant about “Vulgar Marxists”. Add to this the context that ESR has significantly crazified in the last 10/15 years and is now advocating for literal terrorism … yeah (site seems offline, archive, and as with most of his posts the craziest stuff from ESR is in the comments he posts). The reason few people have noticed is probably because ESR has long ago already been relegated to the crazy internet cooks corner by most, and few people have been paying attention to him.

      Besides, the OSI seems to do little more than write blogspam and discus some licensing issues on a mailing list. It’s certainly not significantly involved in the development of Linux as far as I can tell.

    • I’m not aware of any serious efforts to “oust” Linus. The only sources I can find are from techrights.org, and that’s basically the InfoWars of OSS. By and large, people are happy with him as he’s been doing a pretty good job for the last 30 years.

    Lunduke has been harping on about these things for ages, and always just pretends that the context of all of this doesn’t matter and that just “outed == bad == LINUX WILL DIE!!!!” I find it a complete non sequitur.


    Re: “Linux companies”

    • Who cares that IBM “killed off CentOS” (there’s also a bit more nuance to that IMO, but let’s leave that to the side)? There were immediately a bunch of replacements available. Doesn’t sound like “dying” to me.

    • SUSE’s realignment to more “cloud stuff” seems to fit a general trend. Microsoft is doing the same for example: where Windows was once the core product, now e.g. Azure is increasingly seen as its “core product”. For better or worse, the OS in itself has become less important a more “abstracted”.

    • Linux Journal is back, which he conveniently leaves out. LWN is still alive and strong. A single publication running in to a spat trouble strikes me as very little evidence of anything. I don’t see the “onslaught” he’s talking about.


    Re: “Linux complexity”

    Well, the world of computing is more complex than it was in 1992 🤷 It doesn’t seem to me that Linux has it worse than any other mainstream general purpose OS.

    I think his point about maintenance and security are overly simplistic as well; Linux isn’t a monolithic entity where you run every line of code that gets committed to the kernel; it’s probably more useful to see Linux as a sort of “monorepo”.


    Re: “Linux events”

    He is complaining that in-person events were cancelled throughout 2020 and 2021 and that this means “our community is dying” and “in hospice with every known disease on the planet”.

    I kept waiting for him to mention the pandemic.

    He doesn’t. He simple asserts that there are fewer events and that they’re not coming back.

    lol?


    Re: “Fuchsia”

    Yeah, this might replace Linux. We’ll see.

    This is pretty much the only argument that makes any sense: something better will come along and it will displace Linux. It may not even be Fuchsia but something else. I have some ideas in which way things will probably move, but they’re probably wrong. We’ll see what happens.

    Will Linux still be around in 25 years? I don’t think it’s as clear-cut as “Operating systems in the past have come and gone”. Overall, the world of computing is a lot less in its infancy than it was in the 70s and 80s, so it makes sense that systems last longer. There is also the issue there is a lot more software, so compatibility and inertia is more important than ever. You see this with programming languages as well, which seem to have a much longer average longevity than they had in the past.

    It’s may not even be a bad thing if something were to come along and incorporate all the lessons from the last 25 years. Apple didn’t do too bad with its OS X right? But MacOS Classic was kind of horrible, and Linux seems “good enough” for a lot of things. It’s not uncommon that “good enough” blocks “better”, but again, we’ll have to see what happens.

    1. 12

      Regarding the first point, I definitely remember people trying to oust Linus because his behaviour was too brash.

      He went on a small break and came back largely because of those attacks.

      But he’s back now so it largely doesn’t matter, and I think people have stopped attacking him though I’m not certain of that.

      1. 16

        “Oust” is a strong word. I don’t believe I saw anyone credibly suggest that Linus step down from leading kernel development altogether. At most I saw claims that Linus’ interactions in mailing lists etc. was unprofessional, reflected poorly on the Linux project in particular, and could potentially exclude people from wishing to contribute.

        As you said, Linus took these viewpoints to heart and the complaints have died down.

        1. 3

          At most I saw claims that Linus’ interactions in mailing lists etc. was unprofessional, reflected poorly on the Linux project in particular, and could potentially exclude people from wishing to contribute.

          Those tend to be the opening moves of the ousting playbook.

          1. 2

            I try to give people the benefit of the doubt. If they state they’re trying to change Linus’ behavior out of concern for him personally, his legacy, and the health of the kernel development process, I’d accept that, absent any proof of nefarious intent.

        2. 7

          Nobody was trying to oust Linus, just trying to get him to understand that “management by perkele” wasn’t working anymore. Take a read over what he himself posted on LKML on the issue: https://lkml.org/lkml/2018/9/16/167

          1. 6

            FWIW, I think that break did wonders for Linus and was a moment of significant personal growth. The stuff he writes nowadays still has an edge to it, but in a much more mature way. I was recently reading some old Linus rants on the LKML, and I actually cringed each time he wrote that someone “should be retroactively aborted”. I think it’s helpful to think about what happened back then as more of an intervention than an ousting.

            1. 6

              It should also be added that Linus has said things to the effect of “Yeah, I don’t really like this angry temperamental side of my personality either; I wish it was different and I tried to change it and failed, guess it’s just how I am” years earlier already. This wasn’t some sort of magic epiphany moment but just one (large) step in a long process, and neither was it forced upon him by SJW beta cuck feminazi dangerhair Marxists trying to “cancel” him, or some such.

              And while I don’t want to excuse any of his more, ehm, angry behaviour, I also feel that he’s been portrayed a bit unfairly. Some people seem have the impression that he is (or was) some sort of angry madman ranting and raving to everyone because every ridiculously outburst got media attention with a picture of him giving nvidia the finger, because 🍿 Again, not excusing this, but it is a very one-sided and incomplete picture.

              1. 1

                Change is often slow. If over a period of 10 years you manage to reduce non-constructive inflammatory phrasings from being present in 10% of your communications to only 0.1% of communications (while probably also reducing edge cases and improving the general quality), then the outside world will still only get transgressions pointed out to them and observe no change. Even those closer to the fire may draw the same conclusion through confirmation bias.

          2. 9

            You’ve spent a lot of time and effort responding to something that I think (despite the speakers claims otherwise) is basically a clickbait troll. I think that’s laudable but I definitely wouldn’t be spending my own time in this way! It’s a deliberately inflammatory headline and the presentation as a whole is mostly performative incredulousness.

            One thing I would disagree with you about:

            [RMS] probably turned off more people from Free Software than attracted

            I don’t dispute that RMS has variously been unhelpful and has certainly alienated people but I think that still his net attraction to free software has been huge. For me, I first became interested in becoming a computer programmer at all because of his political essays. I don’t think that people who were convinced by his political ideas would change their mind about those ideas even if they later came to dislike him.

            1. 2

              You’ve spent a lot of time and effort responding to something that I think (despite the speakers claims otherwise) is basically a clickbait troll.

              Perhaps; but I’ve seen Lunduke’s stuff around often enough to warrant writing something down, and it wasn’t that time-consuming :-)

              Re: RMS. I don’t want to handwave away your comments, but I’m also a little bit weary of talking about him, so I’ll defer to my post from a few months ago for that, adding to it that “convinced by his political ideas” it’s not an “on/off switch”, and that he mostly turned off people who were broadly sympathetic, but not on all details or with some more nuance, and (strongly) dislike his hard-line no-compromise stance. This was certainly the case for me, and actually quite a few people I know.

              1. 1

                Thanks for linking me to that. Very comprehensive and I think you have me convinced. Having X11 under the GPL would have been an enormous thing for FOSS and I suppose it hadn’t occurred to me that FOSS might have been bigger had Stallman been more charismatic.

              2. 1

                I try to separate RMS the person (whom I have never met) with the ideals he (and the FSF) espouses.

                Many people who are attracted to those ideals can be frustrated that RMS’ personality and communication choices can hinder the wider dissemination of them.

                1. 3

                  I don’t even find it particularly hard (and this is not to say that have taken to a dislike of RMS). There is no shortage of people who I’ve gotten ideas from who I don’t much like.

              3. 5

                There’s also a few “inaccuracies” in the complexity part. Completely omitting the fact that most of the new code comes with drivers. Not sure if the complaint there is “complex hardware is a problem for Linux survival” or … ?

                There’s also the “million lines just in systemd bootloader”. I did not run cloc, but if there’s a million lines just in https://github.com/systemd/systemd/tree/main/src/boot I’ll eat my hat. (Edit: 8.5k lines including comments/whitespace)

                1. 2

                  There’s also a few “inaccuracies” in the complexity part. Completely omitting the fact that most of the new code comes with drivers.

                  Yeah, that’s my feeling as well, but I didn’t feel like doing an examination of the Linux source and what exactly is in those “2 million lines code”; my comment was already long enough 😅 Would be interesting though! Maybe I’ll do it later.

                  As for systemd bootloader, I think he may have been referring to the entire EFI process, but I’d have to go back to see what he said exactly.

                  1. 1

                    I did a cloc on Linux. It’s big, with 20,670,238 LoC in total[1].

                    14,157,243 of those 20MLoC are drivers. Another 993,777 are filesystems. 1,729,484 architecture support code.

                    Ext2 + Ext4 + common filesystem code is 109,636 LoC. The ARM64 support code is just 9,040 LoC. The size of a reasonable Linux system (ARM64, Ext4) would be 3,908,410 LoC[2].

                    Around 4MLoC seems extremely reasonable for the core of a kernel like Linux, in my opinion. And of course there are gonna be drivers and filesystems and additional platforms supported, and each of those things will add a whole bunch of code, but all of that code will be neatly sectioned off in its own area and just isn’t the kind of code which creates a lot of maintainability issues long term.

                    I didn’t bothered to get numbers for exactly where code has gone in recent years, but given that the entire core of a reasonable ARM64 kernel is just 4MLoC, those 2 million LoC per year are definitely going into either drivers, additional CPU architectures, or additional filesystems.

                    [1]: I’ve counted lines labelled as C/C++/headers/Assembly. There are some more code in perl scripts, makefiles, shell scripts, RST files for documentation, etc. But I think it’s fair to say that C, headers and assembly are the meat of the code that actually gets compiled into a Linux kernel. If anything, I over-counted by including a few hundred kLoC of tools and sample code and such.
                    [2]: I arrived at the 4MLoC number by doing (Linux LoC - drivers LoC - architecture LoC - filesystems LoC + ARM64 architecture LoC + Ext2 LoC + Ext4 LoC + filesystem common code LoC).

                  2. 2

                    Completely omitting the fact that most of the new code comes with drivers.

                    Yeah, obviously Intel is going to own the SGX code, and it will have little impact on linux development generally.

                  3. 2

                    He simple asserts that there are fewer events and that they’re not coming back.

                    This is the only point I tentatively agree with—factually at least, not necessarily agree with the fact that it means the community is dying. My feeling is that the number of in-person events has been steadily decreasing in the last decade. I think one of the biggest blows was the 2008 recession though. It may be just my feeling and/or a regional thing, because I don’t have good data.

                    Are there any good catalogs of FOSS events?

                    1. 2

                      Probably; but this seems to be across the board: couchsurfing and meetup.com are also not what they used to be for example. And then there were the forum meetups that I attended, a concept which seems pretty much dead too (just as forums are).

                  4. 18

                    Had to make a brief foray into the world of windows recently…….

                    Oh my, the entire ecosystem feels ……

                    …dirty.

                    Want to read pdf’s? Ha! Ha! Here’s McAfee malware along with it.

                    Want to buy a laptop…. oh my, prepacked with shit.

                    Sigh! I know linux’es problems from the inside better than most…. so don’t even bother to tell me it sucks, point me at something better and say why.

                    But there is a fundamental problem, WinTel abdicated on day 1 any push to create sensible standards for drivers.

                    Thus allowing hardware manufacturers to pile the most unimaginable shit on to the market… so long as they poured in a vast warm steaming pile of drivers (running in kernel mode) that made it “sort of” work.

                    Thus anything better than linux has to come along with millions of lines of complexity, just to get this pile of shit that we currently call PC’s to even boot.

                    The solution will not be (just) a new OS.

                    But new hardware and a clean reabstraction of the hardware interfaces…. and I don’t even begin to see that on the horizon.

                    Fuschia. Yeah, maybe. Don’t care. Like Android, it’s there first and foremost for Google.

                    I’m rather betting on Pinephone and Purism and anyone who follows them.

                    I’m so beyond over some mega corp owning my personal hardware more than I do.

                    And don’t tell me it’s Apple. It’s Apple’s very DNA is to suck as much of the monetary value out of it’s ecosystem it can.

                    So I’ve been with linux since 0.99

                    It’s like my grandfathers axe, my dad replaced the handle and I replaced the head, but I have my grandfathers axe.

                    I’d be fairly surprised if much of the 0.99 code even exists in the kernel anymore.

                    Linux has replaced itself several times over.

                    Will I be with Linux forever?

                    Nah. The POSIX api is b0rked in many places.

                    So what am I betting on? Maybe a fork of the kernel or a twig off the kernel sort of like the RT stuff was.

                    Something that takes what POSIX gives us and does it right.

                    1. 9

                      It’s not just that the Windows ecosystem is dirty, it’s positively unsafe.

                      To pick on one particular vendor - Lenovo shipped Superfish malware on their systems for some time:

                      https://en.wikipedia.org/wiki/Superfish

                      The installation included a universal self-signed certificate authority; the certificate authority allows a man-in-the-middle attack to introduce ads even on encrypted pages. The certificate authority had the same private key across laptops; this allows third-party eavesdroppers to intercept or modify HTTPS secure communications without triggering browser warnings by either extracting the private key or using a self-signed certificate.

                      1. 4

                        And this is impossible to do on Linux? 🤨

                        This has nothing to do with Windows as an OS. You can install dubious software/malware on any system. I can make a “Linux with Superfish” distro just as easy as a Windows image with Superfish.

                        1. 3

                          No, it’s not impossible, and I never made that claim. You’re arguing against a straw man of your own creation.

                          My point was just that this sort of thing is far more common in the Windows ecosystem at the moment. You’re much more likely to wind up with something like Superfish by buying a Windows laptop than a Linux one.

                          Perhaps if preinstalled Linux systems were far more common, and most manufacturers in that sector were in a race to the bottom with ever shrinking margins, it may become common on Linux too.

                      2. 5

                        Yeah, I’ve definitely thought about the idea of something coming to replace Linux. But I don’t trust Fuchsia (even though I don’t think they’re going to shove telemetry in the kernel, it will fundamentally be developed for Google’s needs first and foremost).