1. 14

  2. 4

    That’s a lot of drama. I understand the frustration. I’m happy I moved so much of my DE to TUIs as they run quick and download quick–with the byproduct of getting to avoid some the fragmentation of GUI toolkits. But even outside the GUI, we are lacking color management in Wayland so even deeper the display server situation isn’t good 12 years after its release, and still many communites still on X11 despite it being touted as the future.

    I wonder if Microsoft and Apple would have had similar issues if it were as easy to fork their GUI systems.

    1. 5

      The utter lack of cohesion in GUI systems on Linux is probably one of my least favorite things about using it as a desktop OS.

      1. 5

        I wonder if Microsoft and Apple would have had similar issues if it were as easy to fork their GUI systems.

        Microsoft does. From what I understand (from rumour and old articles, not any personal experience), in the ‘90s there was a lot of tension between the Windows and Office teams, each of which was contributing 40%+ of Microsoft’s total revenue at the time. The Office team didn’t want Windows to add features that made it easy to implement an Office competitor, which limited the built-in GUI features. In addition to this, the developer division regarded Visual Basic / Visual C++ (later Visual Studio) regarded GUI building tools as a differentiating feature.

        This led to a proliferation of third-party toolkits that used Win32 APIs at a similar abstraction layer to X11 and either provided cheaper / easier developer tools than Visual Studio (e.g. Delphi) or provided a richer set of UI components.

        This, in turn, led to a proliferation of different UI appearances and behaviours between different Windows apps.

        In contrast, Apple’s revenue came from hardware sales. The OS existed to sell the hardware and any first-party apps existed to sell the hardware. If another company came along with a better alternative than an Apple app and made a better MacOS version than a Windows version then that was a potential sale for Apple. As a result, they had a strong incentive to make components and services system-wide and reusable between applications. They also had an incentive to write a good set of usability guides and conform to them in all of their first-party software so that things that didn’t conform stood out. This let them sell a consistent experience as a differentiating feature.

        I think one of the biggest problems with open source DEs is that they tend to be written by people who came from Windows.

        1. 1

          Not to mention the constant struggle between DevDiv/Windows teams for their own GUI toolkits and even developer frameworks (if Spolsky is to be believed, IIRC); eternal battle between .NET and COM. (Perhaps they’ve made up, unified on something, and it’s better now. But we still have the fallout of having to choose between rawdog Win32, MFC, ATL, Windows Forms, WPF, and UWP.)

        2. 4

          I’m happy I moved so much of my DE to TUIs

          I think this says a lot about why GUIs on Linux aren’t living up to their potential.

          1. 2

            For Microsoft it depends what you mean by forking GUI. Around w2k/xp it was reasonably simple to make your own desktop environment and there were a few available. They mostly didn’t survive migration to win7, but even now you can find 2 projects that still work. On the GUI as in toolkit side, MS is doing a bit of fragmentation on its own with Win32/WPF/MAUI and there’s a few 3rd party solutions available. And that’s before we even get into “we need to differentiate with this driver settings window so it gets full custom theme”

            While MS avoided some technical issues with its architecture, I find its look and feel fragmentation to be worse than what I get on Linux.

          2. 1

            Well that’s scathing. Sounds like GTK needs a fork.