1. 14
  1.  

  2. 3

    They should have called it “Fragment”!

    1. 2

      There’s precious little ‘new os’ here. The repositories either seem to be empty, or are pristine third party code.

      1. 1

        Please Google: do not kill linux and other UNIXes by making one more OS that take all the market and is not UNIX-compatible! (Why would they do that?!)

        On the other way, an OS that can compile ffmpeg, go, and a built-in terminal may not be that bad.

        1. 4

          Please Google: do not kill linux and other UNIXes by making one more OS that take all the market and is not UNIX-compatible! (Why would they do that?!)

          To actually innovate? (I guess systems software research might not be so irrelevant after all.)

          1. 4

            Does a new OS like this really have to be unix/linux-incompatible? Wouldn’t it make the most sense to try and stay as compatible as possible while innovating in areas that doesn’t impact compatibility? What would one gain from not staying unix/linux-compatible?

            1. 2

              It would be nice to see something that has learned from the past 40 years of computing, for instance.

              1. 2

                Got an example of what the benefits would be? I’m honestly curious.

                1. 3

                  Why are we still pretending that our phones (our phones!) have multiple independent interactive users? Why does the concept of a user need to be conflated with the capabilities a process needs to run? Why is there a root user at all? Why are we still pretending that we need teletype compatibility to enable ad-hoc RPC? Why are we still dealing with the ring 0/1 divide?

            2. 2

              kill linux and other UNIXes

              Without a comparable number of drivers? Don’t hold your breath.

              1. 2

                Drivers matter less nowadays. They’re more complex, but there’s a lot fewer of them you need to support. And if your target is things like phones, then you only need to provide for what’s on the device.

                1. 6

                  Drivers matter less nowadays. They’re more complex, but there’s a lot fewer of them you need to support.

                  Compared to when? I don’t think this is right in any circumstance, and certainly not on a phone.

                  On a PC in the 90’s you’d have to support an Ethernet card, a sound card, a video card, IDE. floppy disk and CD-RW controllers.

                  Now on a PC you have to support all of that (because even if it’s on boarded, you still have to support the controllers for it) plus wireless, understanding the proper way to handle an SSD and of course video card support has gotten orders of magnitude more complex.

                  When you go to a phone you have to add in supporting the cellular radio, GPS, accelerometers, cameras (not just retrieving storage for them but actually controlling them), fingerprint readers and a lot more.

                  I’d argue that today’s phones are the most complicated target an OS has ever had. In the past the OS would never have been expected to support all that itself. It would have only provided the basics and vendors would have provided drivers for anything. But if you’re writing a mobile OS you have to support all that functionality right out of the box, and do it really well with optimized power usage for longer battery life.

            3. 1

              It seems everything has been named after colours. Here is a link to magenta kernel initial documentation: https://fuchsia.googlesource.com/magenta/+/master/docs/index.md