IBM’s systemd attempt to pull the plug on distros using eudev

libgudev is a low level library that bridges connections between udev and graphic stuff (Gobject) and hasn’t been a problem in the past building it with libeudev instead of its own libudev from systemd.

Edition 238 comes out and specifies in code the udev version limit to 251 or higher, with a specific function checking this in the code provided by a similar function in udevd.  Gnome team is the one publishing ibgudev, so you can’t expect to talk sense into them.  The eudev project recently, carried over from the abandoned Gentoo project, is built on 243 edition.

As described by someone on the open issue of eudev the problem is:

packaging-wise, it’s trivial to just patch such a check (in the meson.build dependency for libudev, inside libgudev’s build; i’ve had to do that multiple times when things declare some high version but don’t actually need it at all).

however, it’s not actually that trivial in this case, because the actual function they bumped for is missing:

../gudev/gudevdevice.c:146:12: error: implicit declaration of function 'udev_device_get_current_tags_list_entry'; did you mean 'udev_device_get_tags_list_entr'? [-Werror=implicit-function-declaration]   146 |   for (l = udev_device_get_current_tags_list_entry (device->priv->udevice); l != NULL; l = udev_list_entry_get_next (l))       |            ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 

this was added in systemd/systemd@3b684be, so 247 was the first version with it. there’s no further requirements in this case..

i don’t know how easy it would be to implement this one function.

 

Being a low level library a ton of graphical based software are directly or indirectly linked to it and this may cause several problems to distros being partially dependent for software on a systemd distro, such as Arch or others (later).   If eudev team take too long to address this issue and bump their version up to 251 and above, connectivity will be lost, software will begin breaking based on wrong edition of libgudev, and things generally will begin to fall apart.

Distros that have switched from eudev and build udev/libudev straight out of the systemd repository will face no issue.  So “why is it that this team (gnome, systemd, …pulse/pipew…) want to force people to switch to systemd’s udevd instead of an alternative?  Or is the whole change another attempt to force everyone to abide by their rules and defeat any available choice?

This is why we keep saying that solutions and methods such as what artix uses are only on a legality systemd-free, by altering pid1 from systemd to something else, otherwise udevd elogind and many of their libraries are all there and essential to their system.  Does something like OpenRC pose as an improvement to systemd?

The core question here, again and again, is not what this corporation does to manhandle and control FOSS, is that people don’t see the immoral and unethical ways they employee to control everything and everyone.  Totally against the spirit of gnu, foss, FSF, unix, whatever you want to call it.  They are driving everything to a uniformity of a single system.  Linus is whistling apathetically like if it was none of his business.

So let us see whether the eudev team is up to the task in saving linux for us.  Systemd is up to 253, eudev is back on 243, ten major editions behind.

 

 

 

 

16 thoughts on “IBM’s systemd attempt to pull the plug on distros using eudev

    • I am looking up the date, in which phase is Hyperbola now, moving to BSD or moving back?

      I may have mentioned this before, in case you didn’t see it, for as long as we know linux through linux.org, new editions and especially new -lts kernels branded as such have had a linearly incremental deadline. In the past couple of years all deadlines seem to converge to 12/2026. No current kernel seems to be able to exceed this date,4.9 only recently was terminated. So it may not be just Hyperbola but we may all be looking for a kernel in 3.5 years. So keep stocking up on tarballs for the future of your favorite ones.

      I am speculating Linus is going on early retirement and the remaining gang will drastically change the character, maybe to a commercial /non-free product although open-source. They can also drop all hw that is 5+ year old, to agree with industry recycling tendencies.

      My newest machine exceeds a decade .. and it is perfectly humming compiling and linking daily.

      Like

  1. Hyperbola is moving to a forked OpenBSD base, which is difficult as much legacy code is not sufficiently free for the purposes of Hyperbola.

    OpenBSD will not help, for reasons explained long ago by Cynwulf on here.

    Like

    • Cynwulf is a psychopath 2 faced snake I would like to meet one day and give “it” the anti-fascist treatment he deserves. So please don’t mention this snake’s name here.
      When cynwulf could not rationally argue against mobinmob on his claim that BSD scripts were better than runit or s6, while I failed to participate, he turned to say he wasn’t “protected”, then turned into OpenBSD forums to say this is a political propaganda site masqueraded as a linux site. This pathology of turning against people because one “can not win an argument” is just as bad as claiming to be a fascist. One must win an argument, even though they are not rationally able to convince anyone, they still want to win. And in winning they can spread myths and lies about the other party. Unfortunately for this psychopath and the tendency we have had since day one, we do not delete messages we don’t like, they are still here for people to judge. Everytime Cynwulf lacked an argument to support his claim he would shift the topic and attack the other party participating in the debate.

      Does that make OpenBSD worse, no, but Cynwulves can float in their OpenNess.

      Somehow Hyperbola thought they can transfer all work done in GPL into a BSD licensed system, not realizing they have to violate either the one or the other in doing so. Since then they insisted in projecting a future OpenBSD system which AFAIK never materialized while allowing the linux system to fail. Then returned to maintain the linux part, while shoe-horning some BSD stuff into it.

      This whole attempt had Hyperbola users dizzy for a while. The excuse was they couldn’t keep up cleaning the kernel enough to make it -free but they returned and published a few more versions of kernels they had previously said couldn’t be cleaned. They are almost as bad as Joborun in this respect, they think aloud.

      Like

      • Depending on Hyperbola: This is not the full story you are telling here. We have never thought to transfer from BSD to GPL. We have always said that we want to rewrite code and license this under the GPL. As HyperbolaBSD is planned as new BSD-descendant operating-system, not just an OpenBSD-fork, this is our development going on. Also we have not “shoe-horned some BSD-stuff into”: Hyperbola GNU/Linux-libre is on the point for transitioning from GNU/Linux to HyperbolaBSD. We have also not returned, we maintain the GNU/Linux-system as long as we think possible.

        Your last paragraph is not clear for me, but speaking of: GNU/Linux-libre is not a hard fork like HyperBK (our BSD-kernel). So you miss the point here. If you like to discuss that matter more direct: You can always get in touch with us.

        throgh from Hyperbola

        Like

  2. Hello,

    https://github.com/eudev-project/eudev

    “This is a project started by Gentoo developers and testing was initially being done mostly on OpenRC. We welcome contribution from others using a variety of system initializations to ensure eudev remains system initialization and distribution neutral. On 2021-08-20 Gentoo decided to abandon eudev and a new project was established on 2021-09-14 by Alpine, Devuan and Gentoo contributors (alphabetical order).”

    There’s no problems, only solutions, but if there’s no solution, so there’s no problem…

    Like

    • Although I can understand what you mean by it, in absolute terms a problem will continue to be a problem as longs as there will be no solution.

      So now we have a problem, it is called insufficient eudev development, to the point that antiX, obarun, joborun, have stumbled to abandon their eudev support.

      Like

    • To add – only ‘solution’ at the moment it to pin 237 version before upgrading.
      If users have already upgraded, then they will need to downgrade to 272 versions of libgudev

      Like

      • I don’t know which is the problem or the solution.
        In arch(pacman) you don’t pin versions of sw, there is a simplistic but effective way of hierarchical ordering of repositories, and so joborun repos can be on top of obarun and obarun on top of arch and the repos below can release anything they want but an upgrade will not take place unless a higher numbered (ver/rel) is released by the top repository. The issue though is with dependents and sub-dependents being released by the mothership where you don’t provide alternative builds. Whether you can keep up building anything affected and add them to prevent users’ breaking applications.

        So what is the issue here. Is this feature earlier udev didn’t have that is needed that requires a later version or just a check on version? Is this feature something meaningful or something fabricated to shake off slow-following-forks of the original? Despite of how much we like to hate conspiracy theories, there have been many conspiracy theories which were found to be correct theories lacking support at their early stage.

        From the point of view of domination, among complete and extensive linux distros, void and gentoo were major Galatian/Galician villages resisting the empire. So udev is a significant asset they need conquered and controlled. Hmmm…!!! Reverse engineer the monster (Hydra Capitalista) and you may draw interesting conclusions.

        Like

        • The problem is that eudev is no longer being maintained it seems.
          They are so far behind udev/systemd (as you point out in this article) that catching up and removing the offending systemd parts will take a lot of time.
          Since libudev 238 breaks keyboard and mouse usage (and probably others) a short term solution is required so users can physically use their computers/laptops.
          antiX has uploaded libgudev_238-100~antix1 to its repos that, at least for a while, will keep users on a working device.
          Any previous pining/holding should be reversed before upgrading the antiX libgudev_238-100~antix1 deb

          Like

  3. I just updated Devuan Unstable’s packages and eudev completely fell apart, freezing even the TTY when Xorg is loaded. I conveniently found this post in a search for answers. Oh the horror…

    Like

  4. Some feedback from one eudev dev. Make of it what you wish.

    There is no need to port everything, just the new API and its dependencies.

    I am traveling and the best I can do now is to review/merge a PR (in case some fine folks make one).

    I have no idea about the scope of the change – it may be very small or really huge…

    Like

    • I don’t know which is the problem or the solution.
      In arch(pacman) you don’t pin versions of sw, there is a simplistic but effective way of hierarchical ordering of repositories, and so joborun repos can be on top of obarun and obarun on top of arch and the repos below can release anything they want but an upgrade will not take place unless a higher numbered (ver/rel) is released by the top repository. The issue though is with dependents and sub-dependents being released by the mothership where you don’t provide alternative builds. Whether you can keep up building anything affected and add them to prevent users’ breaking applications.

      What if upower and other direct dependencies are to be built/rebuilt and pinned and hope the rest last long enough to have a fresher copy of eudev.

      So what is the issue here. Is this feature earlier udev didn’t have that is needed that requires a later version or just a check on version? Is this feature something meaningful or something fabricated to shake off slow-following-forks of the original? Despite of how much we like to hate conspiracy theories, there have been many conspiracy theories which were found to be correct theories lacking support at their early stage.

      From the point of view of domination, among complete and extensive linux distros, void and gentoo were major Galatian/Galician villages resisting the empire. So udev is a significant asset they need conquered and controlled. Hmmm…!!! Reverse engineer the monster (Hydra Capitalista) and you may draw interesting conclusions.

      Unless you think it is all coincidental to development needs.

      Like

  5. Microsoft hiring Lennart Poetering and IBM’s acquisition of Red Hat may not be as random or as haphazard as they first appear.

    This is just my opinion:

    TPTSNB want to control everything, from the top down. Linux, previously, had been about options and freedom; however, with useful software being tied directly to systemd, a project “they” own and control, it’s only a matter of time before non-systemd distros will be nothing but curses and ascii-based applications with little to no online functionality.

    To be fair, I have a bleak outlook on life in general, but, that’s only do to my work experience across a couple of continents.

    Like

    • The amount of coding that is used for windows-ms (the wo/man-hours available) can only be a small fraction of those who work on open/free code globally. So logically there is no way in hell that Apple/MS can catch up to the speed-train called FOSS. Even if those guys are paid and the rest do it for hobby (potential employment/commercial activity) … they are midgets. So undoubtedly they steal, copy/past code from FOSS. Since IBM (MS nemesis who lost DOS out of their hands for peanuts) has systemd and can control its use and development (my guess is still that IBM had been hiding behind RH since day 1, funding it secretly with consulting sub-contracts passed on) MS must have a response, a systemd from scratch by the same maestro who orchistrated the original. He now knows the loopholes of dead-end development, and can redesign from 0 something with greater potential, to drive MS internally.

      I am also speculating the kernel has an expiration date itself, and it will stop being FOSS in the near future, as soon as Torvalds announces retirement (millions can allow someone to retire young, and he is not even that young anymore). Watch the release schedule closely, it seems lights are going out in Dec/26 Then we will be stuck with refining forks of the last kernel issued for decades, if hw cooperates.

      Like

If your comment is considered off-topic a new topic will be created with your comment to continue a different discussion. This community is based on open and free communication, meaning we must all respect all in minimizing the exercise of freedom to disrupt such communication. Feel free to post what you think but keep in mind the subject matter discussed. It is just as easy to start a new topic as it is to dilute the content of an existing discussion.

This site uses Akismet to reduce spam. Learn how your comment data is processed.