Announcement

Collapse
No announcement yet.

A Vast Majority Of Linux's Input Improvements Are Developed By One Individual

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #21
    Originally posted by jo-erlend View Post

    Are you the same person who's repeatedly claiming that GPLv3 is a proprietary software license here, or did you just copy?
    Neither.

    Comment


    • #22
      I've been actively reading and using libinput the last month writing a Tablet config KCM for it on X and Wayland. And even just doing this I'm itching and aching to get back to Rust - I'm passing raw pointers between QML and C++ and hating every moment of it. I have a half dozen reinterpret casts to be able to interact with the XInput API because its all raw char* buffers and page long function signatures with the best documentation sometimes being a man page. But often its just "theres this named constant used in this one example, there might be another constant in some X header that does something similar that I'm looking for" then an hour of Googling and reading C headers buried in preprocessor barf you might find what you are looking for amongst a hundred other undocumented unmentioned variables in XCB or Xlib or XI or XInput or Xatom or one of the other hundred header files. That, versus actual search in generated documentation I have locally, is absolute night and day.

      I fear that might happen more and more as the next generation of programming languages take off and all the foundations of the Linux desktop rot without new blood. Because the new blood is attracted to more well designed languages than C and C++. I basically started using KDE exclusively almost a decade ago because after having dealt with gobject C code in uni I never wanted to ever see it again. I remember back in ~2014 I was considering trying to contribute to the dying Telepathy project but once I realized it was all internally that same horrifying gobject C gobblygook I walked away.
      Last edited by zanny; 10-16-2019, 03:00 PM.

      Comment


      • #23
        Originally posted by Britoid View Post

        Neither.
        Ok. So out of curiosity, what makes you feel that software licensed under GPLv3 is proprietary?

        Comment


        • #24
          Originally posted by schmidtbag View Post
          It's all a matter of perspective. It's like asking who was victorious in a war - there's no truly right answer, it all depends on where you stand in the matter. So, if you have a gripe with RH, you're going to think they're to blame for fragmentation, and there is some validity in you thinking that. But like I said before, everyone has their own idea of what the best thing is. There is no such thing as a universally agreed upon solution in Linux, and that's why forked distros happen. In other words, fragmentation occurs when people demand change. That being said, I feel Red Hat isn't at fault here because they're one of the very few leading organizations in Linux development that remains relatively consistent, for better or worse. In the corporate world, consistency and reliability is often more important than performance or efficiency. Of course, with Linux being open-source, RH can't win. There's nothing they can do to prevent fragmentation, but at the very least, they can maintain some stability in the corporate Linux world.
          Meanwhile, look at how Windows works. Most of Windows is crap. A lot of people will install bloated 3rd party tools just to override how crappy the built-in features are. MS could easily adapt, but they don't, because that's the only way they can retain compatibility and consistency (which also includes familiarity), and that's really the only reason people use Windows. This is also why Windows 8 and Vista were hated so much - they both broke consistency.
          Anyway, the reason Canonical gets so much flak is because they reject other people's ideas while also forcing others upon their own. Take Mir, Upstart, and Unity for example - each of those were a highly unnecessary distraction, and those resources were better off spent elsewhere. So, they're just as stubborn as RH except instead of just simply turning down better alternatives, they make yet another one.

          Canonical used to be pragmatic; they aren't at all anymore, and haven't been since maybe 2011. That being said, I do thank them for making Linux a more user-friendly experience, because Ubuntu is the reason I got as far as I did with Linux. I first started looking into Linux around 2006 but Ubuntu was the only distro that didn't feel like I was learning how to swim in the middle of the ocean. Today, Ubuntu is too fragile and inconsistent. It's not a bad distro, but if I were a beginner, the amount of deviations they do from other distros (and previous versions of Ubuntu...) would be extremely offputting to me.
          Upstart came years before SystemD. Mir was indeed a NIH. Unity was born out of the disaster that was the initial release of Gnome 3. And the way I see it, they showed the way to Gnome developers, at a time where they thought a desktop environment didn't needed to be shut down or a window to be resized. And lets not forget that things were so bad that Canonical wasn't alone in creating a fork of Gnome. And Red Hat stayed away of the mess for as long as they could, until the dust had settle down and Gnome 3 was something usable.

          Comment


          • #25
            Originally posted by [email protected] View Post
            Upstart came years before SystemD. Mir was indeed a NIH. Unity was born out of the disaster that was the initial release of Gnome 3. And the way I see it, they showed the way to Gnome developers, at a time where they thought a desktop environment didn't needed to be shut down or a window to be resized. And lets not forget that things were so bad that Canonical wasn't alone in creating a fork of Gnome. And Red Hat stayed away of the mess for as long as they could, until the dust had settle down and Gnome 3 was something usable.
            Hmm, for whatever reason I thought upstart came after systemd. My bad.
            Regardless of how bad Gnome 3 was, Unity (which also was pretty broken to start with) was still heavily dependent on it. Rather than make their own fork, Canonical could've just helped Gnome 3 get better and speed up the process. Remember: the reason people gripe about Canonical is because of how many times they fork things. It's not always something as big as Unity or Mir either.
            Anyway, I'm aware there were other forks, but to my knowledge, most of them were basically just mashups of different environments. Unity was far more ambitious, and the greatest issue is Canonical actually had the resources to fix Gnome 3. Instead, they just decided to do things their own way. Same applies to Mir.

            Comment


            • #26
              Originally posted by schmidtbag View Post
              He seems to be doing a good job, and if anything does happen to him, I'm sure Red Hat can find a replacement.
              it doesn't work that way. redhat can hire someone who has made himself replacement, that's all. redhat can't tell someone to become replacement

              Comment


              • #27
                Originally posted by Britoid View Post

                With no Red Hat and its employees, half of the current Linux userspace wouldn't exist in its current form (gnome, wayland, systemd, dbus, colord, NetworkManager, polkit, sssd, packagekit, kvm).

                I hope IBM keeps the Red Hat technique of promoting independent open source projects of which then you build commercial products on top.
                As much as I prefer to work with Redhat...

                Having those things 'thanos-snapped' out of existance would be an overall improvement for enterprise/server Linux (eg, for headless VMs that never run a UI other than bash)....

                It would almost be better if 'Desktop Linux' forked off the way the BSDs split up, such that the next 'good idea' for trying to create a competitive end-user OS wouldn't ripple through the server ecosystem where the change in question is unneeded if-not harmful.

                Comment


                • #28
                  Originally posted by Britoid View Post
                  With no Red Hat and its employees, half of the current Linux userspace wouldn't exist in its current form (gnome, wayland, systemd, dbus, colord, NetworkManager, polkit, sssd, packagekit, kvm).
                  And about half of that wouldn't actually be a loss - just like half of the xinput changes wouldn't be.

                  Hutterer is a perfect example of the Red Hat mentality: this is someone who broke the hell out of mouse acceleration in X, unilaterally. Don't get me wrong: it WAS already broken, no question, and the code was terrible - but the new LOGIC isn't really any better, and there was no discussion of it and no attempt to get a result that was even remotely close to what it had been for over a decade. Just "here's some random behavior that I pulled out of my ass, here you go", with the resultant code being utterly unusable with anything other than garbage-tier mice, because the acceleration on anything with a sample rate faster than about 50ms or with more than 50 CPI is INSANE.

                  In fairness, he did at least add a way to hack xinput to try and deal with that, but it's clunky as hell. To this day, I have multiple machines where the mouse is literally unusable without "xinput --prop-set etc etc etc" hacked into /etc/.bashrc of all places.

                  As someone who's had to suffer through writing xinput code, I can certainly agree with zanny that the documentation is borderline useless, and the API is absolutely terrible. The only place half of it is actually "documented" is in his @#$%ing BLOG, fgs.

                  So yeah, it's not great that he's the only xinput developer, and not just because that's not a good place for a crucial system to ever be in. But that's the Red Hat way of doing things: produce something in isolation, and say "Here, deal with it". They're no different than Canonical in that regard, and often much worse. So while I sympathise with Hutterer's predicament, you don't get to complain about how you're on your own when you (or in this case, I assume "your employer") decide that's how you're going to build the thing in the first place.

                  Comment


                  • #29
                    The guy needs some dang help with laptops for testing. He doesn't have nearly enough of them as far as I can tell. libinput being nearly a one-man show is kind of ridiculous.

                    Comment


                    • #30
                      Scary stuff. Since it's open source and no one else is apparently interested...let's hope nothing happens to him.

                      Comment

                      Working...
                      X