Announcement

Collapse
No announcement yet.

Systemd 236 Is Being Prepped For Release This Month With Many Changes

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

  • #21
    Originally posted by jpg44 View Post
    This is my honestly held and sincere view of the systemd matter: I never got the hatred of systemd. For me, its worked absolutely fine. Its a huge improvement that has increased the amount of control over initialization for the user. The declarative style service files are actually much easier to read and understand the shell scripts. The big seller for me is the fact that you can still do SysV init files to your hearts content. I have tested SysV init files with systemd and all works perfectly well. If people do not like the dependency based start up, anyone who wants to can have their service started from a SysV init file and disable the systemd service file. So given that systemd does not remove any functionality and can run SysV init files, and all functionality it adds is an addition to the traditional SysV functionality, it doesnt take away any choice or freedom from anyone.

    It actually increases freedom because you can write DBUS based init code that can include any kind of heuristics you want to respond to any series of events. Because of the Bus based design of systemd, its much easier to monitor for system events and events generated by other programs. The system is actually far more modular SysV init, because you can write your own init daemon which can monitor for any other events on the system.

    So given that systemd includes all previous functionality, is backwards compatible, those who oppose it seem to think that they need to be able to deny people the ability to use its features, because systemd doesn't actually deny anyone the ability to use SysV type init for their services.

    Remember that Ubuntu had Upstart for years which was basically the same thing as systemd, when Ubuntu went to systemd, it wasnt anything new and was an improvement since it standardized on what the other distros have been doing.

    For people who do not like the systemd type init files. The fact is, you don't have to use them, just do a SysV init file and systemd will use those.

    You try to explain any of this to people, and it flies right over their head and they absolutely refuse to listen. Its just pointless to try to debate with these people. They refuse to listen adn they have this idea in their mind and they refuse to see the facts that clearly contradict it.

    Then they will repeat the same old debunked myths about it that have been debunked over and over again, and which you have just explained to them are not true. Like it being "monolithic", its not, its more decentralized that SysV init was because it is a client server bus based design that is totally decentralized and you can write your own daemon in any language you want to start your services. Or that it takes away freedom, which is the opposite of what is true since you actually have more freedom and control since you can start services on any prerequisite series of system events by watching DBUS, according to whatever heuristic logic you need, if you dont like the service files you are free to use procedural code in your own daemon that can watch DBUS for whatever events you want to respond to. It also does not prevent you from using a cron file to start services (systemd offers similar functionality but you are not required to use it).

    People who try to attack Poettering by suggesting that its a power grab ignore that systemd is more configurable than sysv init is. I honestly think that Poettering has and intended to increase user freedom and and flexibility to users because of the decentralized bus design of systemd.
    Exactly , Perfect, Amen and so on.... I quoted the entire post since it is worth reading again!

    http://www.dirtcellar.net

    Comment


    • #22
      Originally posted by WhackJob
      Nobody is forcing Poettering to use Linux, he can switch to BSD if he doesn't like it.
      Same to you.

      Comment


      • #23
        Originally posted by debianxfce View Post
        use alsa, gstreamer and jackd ,... use sysv or openrc)
        they did not teach you to reuse software and not re-inveting the wheel? you could choose only one of those

        Comment


        • #24
          Originally posted by L_A_G View Post
          The fact that something needed to be made doesn't mean you can't criticize the botched implementation of it
          but the fact that you did not make anything worthwhile does mean that nobody is going to take your criticism seriously
          Last edited by pal666; 01 December 2017, 09:16 PM.

          Comment


          • #25
            Originally posted by InsideJob View Post
            I still think the kernel needs a "feature" than just makes systemd no longer work.
            your bsd kernel had that feature from beginning

            Comment


            • #26
              Originally posted by debianxfce View Post
              Poettering and other redhat idots do not realise that some embedded devices can use old 3.18 android based kernels
              I'm sorry, but if you've got locked on ancient kernel, it's hardly RH fault. RH could be blamed for many things, but surely not this one. Whatever, ancient vendor kernel suxx. Linux is only fun when it happens to be mainline.

              Then, neither sysv init nor openrc, etc would take care of truckload of system-level challenges like systemd does. Do it yourself, reinventing your fucking squared wheel 22nd time in a row. Then, alsa. C'mon, its API is awful and leaves a lot to be desired. Jack is inclined on professional use cases. Hopefully it explains why so many apps are using pulseaudio to play some sounds here and there.
              Last edited by SystemCrasher; 01 December 2017, 10:50 PM.

              Comment


              • #27
                Originally posted by starshipeleven View Post
                Armlogic is a shit because their hardware decoding accelerator does not have an open driver, among other things like audio and some HDMI issues that had to be worked around.
                If I read the linux-meson matrix correctly, some boards like Odroid C1 don't even have SMP (multi-core) enabled yet unless you use kernel 4.15-rc1. Which kind of shows how shitty the support is. The board is already old as heck.

                Comment


                • #28
                  Originally posted by InsideJob View Post
                  I still think the kernel needs a "feature" than just makes systemd no longer work.
                  At this point you'll take your stand straight against mr. Torvalds himself, who would shout loud and clearly they do not break user mode stuff. Have fun challenging him and don't forget to post links if this happens. I'm just curious if you'll manage to earn some strong, colorful objections from mr. Torvalds.

                  Nobody is forcing Poettering to use Linux, he can switch to BSD if he doesn't like it. What goes around comes around.
                  Poettering seems to be okay with Linux, especially when it runs his own PID 1 thing. So if you do not like particular distro, feel free to change it or even switch to BSD or whatever you like. It is up to distro maintainers to choose what they would like to maintain. Maintaining just some small text file instead of fairly big and troublesome init scripts seems to be appealing idea. Not to mention systemd allows to use busload of system-level features easily. Not something one could easily do with shell scripts, due to fragile and complicated nature of system-level calls and fact most of them require elevated rights. So merely putting service into container while switching user and andjusting priority could prove to be quite a challenge if performed from shell scripts.

                  Comment


                  • #29
                    Originally posted by debianxfce View Post
                    Poettering and other redhat idots do not realise that some embedded devices can use old 3.18 android based kernels only if you want all the hw bells and whistles working. Making systemd bug reports is useless. You can run Debian testing without systemd running luckily. I am jealous to distributions that use sysv or openrc.
                    http://without-systemd.org/wiki/inde...Debian_Stretch
                    init=/lib/sysvinit/init in the kernel command line to prevent systemd to work.

                    And that's exactly the kind of crap you have to deal with when using hardware without open source drivers. Try weird workarounds, and all sorts of antics because the drivers you want will only work with kernel X.YZ and you have no way to make them work with newer versions of Linux kernel and userspace software. This is why attempting to build an "open" phone/tablet using the current common hardware (Snapdragon SoC, Exynos etc.) is a huge waste of time - seriously people, just use a system with Intel GPU.

                    Seriously Canonical - you could've atleast launched Ubuntu tablet using Intel GPU which would've saved you a huge amount of time and pain, allowed you to use pretty much same software stack as desktop distro, and use a really solid stable open source GPU driver with all the bells and whistles. Or even an AMD GPU (which are way more powerful than the Intel GPUs).

                    And the KDE tablet - using an Intel GPU would've been so much better instead of actually believing you would get non-existent open source drivers for hardware.

                    Intel already walked the talk and continue to do so. If anyone who's planning to launch tablets/phones with open source software is listening, please please please use an Intel GPU so that we can actually have an open and functional system - yes I get that it's not going to be perfect (Skylake+ GPU requires closed source firmware, Intel ME and all that) but atleast the operating system and other parts will actually be open source. Open source firmware and open hardware are useless if you have to run closed source drivers to actually use the hardware.

                    Comment


                    • #30
                      Originally posted by starshipeleven View Post
                      Yeah, it's yet another case of an old program doing dodgy things that gets blocked by systemd.

                      How bad are systemd devs for asking downstream to fix their unsafe shit. waah waah waah.
                      They are not asking anybody, they are breaking shit. I would be much more inclined to accept this if they added functionality to journald that spews "XYZ is deprecated and will break in future version of systemd" for 5 versions before they actually break anything. Breaking userspace left and right and telling users to go fuck themselves if something breaks _shouldn't_ be acceptable approach. Fortunately most of this crap gets caught by distro package maintainers long before it hits users.
                      Unless you are running software in your business that you bought 5 years ago, don't have source codes level access to and to prevent this you just never update your system(d). #neverhappened

                      Comment

                      Working...
                      X