Poettering said...
|
Author | Content |
---|---|
tuppp Aug 29, 2015 10:58 PM EDT |
"'su' is really a broken concept. It will given you kind of a shell, and it's fine to use it for that, but it's not a full login, and shouldn't be mistaken for one." su always seems to work perfectly for me. I wonder if Poettering has ever tried "su root" or "su <user>" or "su -". |
penguinist Aug 29, 2015 11:25 PM EDT |
systemd is turning into a cancer! Somebody needs to tell Poettering to stop it. |
arm Aug 29, 2015 11:35 PM EDT |
It's only the tip of the iceberg. ls, cat, and other standard tools will be next. Poettering wont be happy until Systemd is the operating system. |
nmset Aug 30, 2015 5:23 AM EDT |
This does not mean su will ever disappear. There has long been systemd-networkd, yet other network managers are still shipped in distros and no one is required to use systemd _additional_tools_. |
750 Aug 30, 2015 6:31 AM EDT |
_tuppp, su is broken because it breaks systemd seems to be the unstated logic. nmset, i have already seen projects list systemd as part of their compile dependencies. And if this keeps on going i suspect i will see more. |
jdixon Aug 30, 2015 6:38 AM EDT |
> su is broken because it breaks systemd seems to be the unstated logic. I hadn't considered that, but it does sound that way, doesn't it. > i have already seen projects list systemd as part of their compile dependencies. As I've noted in earlier threads on this subject, if Slackware ever has to go with systemd, it may be time to look at one of the BSD projects. |
penguinist Aug 30, 2015 8:31 AM EDT |
With Linux it is all about choice: Gnu/Linux Systemd/Linux Android/Linux Three variants all sitting on top of the same kernel. |
mbaehrlxer Aug 30, 2015 11:58 AM EDT |
well, my first thought too was WAAT? but "machine shell" does something different than su by "creating su(1)-like privileged sessions, that are fully isolated from the original session", which su does not. so running "machine shell" is more equivalent to running "ssh root@localhost", which, if you block root logins, can't be used to get a new root shell. greetings, eMBee. |
GDStewart Aug 30, 2015 1:01 PM EDT |
"systemd is turning into a cancer! Somebody needs to tell Poettering to stop it." Unfortunately the only single person that can stop him (Linus) is quoted as saying that he doesn't see anything wrong with it and doesn't understand all the "hate" directed toward it. Major distributions could also stop it but so far they are almost all headed in the exact opposite direction. We are royally screwed! Personally, I don't give a rat's behind about systemd but if other people want to use it fine. I just want to be able to have a fully functional Debian distribution that does not in any way depend on it (no little helper programs to translate what looks like "normal" Linux init/network/logging/su/etc., etc., etc., etc., ... to systemd and back). |
750 Aug 30, 2015 1:19 PM EDT |
GDStewart, sadly that Torvalds line seems to cover the whole of systemd when it was basically aimed at systemd-as-init. That may be one of the more insidious issues related to debating systemd, that the init process and the larger project carries the same name. Thus you have all kind of confusion that proponents can use as cheap leverage to make people look dumb. |
jdixon Aug 30, 2015 1:43 PM EDT |
> With Linux it is all about choice: Not if Poettering has anything to say about it. :( |
cybertao Aug 30, 2015 4:16 PM EDT |
I'm happily using my GNU/Linux OS that utilizes Systemd. The more I learn about and use it, the more I like and appreciate it. |
GDStewart Aug 30, 2015 5:08 PM EDT |
"GDStewart, sadly that Torvalds line seems to cover the whole of systemd when it was basically aimed at systemd-as-init." Not quite sure what you mean here but Linus said that very recently, within the last few months so it was well after it became the "Monster that Ate Linux" ;) "I'm happily using my GNU/Linux OS that utilizes Systemd. The more I learn about and use it, the more I like and appreciate it." Great, as long as I have the choice to NOT use it with my favorite distribution and still have a fully functional Linux system I have no problem with it at all. But that is clearly not the final outcome Pottering is aiming for and that is why I oppose it. |
cybertao Aug 30, 2015 5:20 PM EDT |
Do you imagine Pottering wearing an evil looking mask as he maliciously plots to destroy the GNU/Linux ecosystem? |
jdixon Aug 30, 2015 6:01 PM EDT |
> Do you imagine Pottering wearing an evil looking mask as he maliciously plots to destroy the GNU/Linux ecosystem? "Never attribute to malice that which is adequately explained by stupidity." Or in this case, willful ignorance. |
gus3 Aug 30, 2015 6:12 PM EDT |
I dunno. Attributing it to malice might be more entertaining. |
AwesomeTux Aug 31, 2015 1:03 AM EDT |
GDStewart wrote:"Unfortunately the only single person that can stop him (Linus)" Linus has no control over non-kernel components. Linux starts SystemD then steps aside, letting SystemD take control. Thinking Linus has any control over SystemD is like thinking GNU has control over Linux simply because GRUB is responsible for starting Linux. However, with that logic GRUB actually does have a certain amount of control over Linux, since it does control what mode-setting and modules Linux uses and initializes multiple other things Linux will require, the same is not true of Linux for SystemD. In this particular case "su" is a GNU package, so if anyone could stop him it's the GNU Project. But SystemD isn't actually using the code for "su", so, all they could do is try to convince him not to incorporate a replacement for "su", or make "su" do what SystemD needs. Neither is likely. |
mbaehrlxer Aug 31, 2015 2:06 AM EDT |
the only way linus could stop systemd would be by writing his own replacement for init. people are going to jump to it in droves and will quickly fill in the pieces needed to make systemd fully replaceable. greetings, eMBee. |
AwesomeTux Aug 31, 2015 2:44 AM EDT |
As I understand, it's been quite a while since Linus has written any code. For Linux, Git, or otherwise. But if he did, that would probably be a good solution, assuming he understands any of the facets of the userland portions of an operating system. In my opinion, I'd rather see an init system from GNU, perhaps as a subsystem of GRUB, or from freedesktop.org as a lower system of Wayland, so it could control mode-setting, etc. This way at least one of the groups who already develop the userland parts of the operating system would be developing the init system. Though, I personally use SystemD, and have had no problems with it, either. |
arm Aug 31, 2015 3:26 AM EDT |
Quoting:In my opinion, I'd rather see an init system from GNU, perhaps as a subsystem of GRUB, or from freedesktop.org as a lower system of Wayland Make the init system a subsystem of the bootloader or a display server. Yes, that makes sense doesn't it. oO |
AwesomeTux Aug 31, 2015 3:59 AM EDT |
arm wrote:Make the init system a subsystem of the bootloader or a display server. Yes, that makes sense doesn't it. oO Not a "subsystem" to those, those would be subsystems of it. But yes. The bootloader starts up the initial operating system processes, so it would make sense for it to also include another process for continuing the boot procedure. As for "display server", yes, the boot process needs to start the graphic user interface and adjust the resolution of the TTY and GUI, so it too makes sense that the graphic server that is inevitably going to take control should be responsible for starting itself, or at least share developers. In the case of Wayland, since it's not a "display server" in the traditional sense, it's a "protocol between a compositor and its clients" therefore I don't think it would be as cumbersome or as odd as you might think it would be. Considering there's already a process in the boot sequence (GDM) for starting the GUI, it's easy to imagine that process being the init system, or at least part of it. To be clear, I said I'd prefer GNU or freedesktop.org create an init system. I use GRUB and Wayland as examples, to say the developers of GNU and freedesktop.org have already written software both in the boot process and essential for the operating system, and both are already familiar with userland and the boot process. |
GDStewart Aug 31, 2015 8:29 AM EDT |
"Linus has no control over non-kernel components.:" I never said he had "control", however he does have a great deal of influence in the Linux community and his approval of systemd is only encouraging Poettering to continue feeding the monster. "In this particular case "su" is a GNU package, so if anyone could stop him it's the GNU Project. But SystemD isn't actually using the code for "su", so, all they could do is try to convince him not to incorporate a replacement for "su", or make "su" do what SystemD needs. Neither is likely." I don't care if systemd adds some replacement for su or not. I don't care if systemd sucks in every GNU utility on Earth. I just want to have a choice to use it or not use it and not have that choice severely limit which distributions I can use. "the only way linus could stop systemd would be by writing his own replacement for init" He doesn't have to write anything, the original Sys V and several other inits already exist. |
AwesomeTux Aug 31, 2015 9:34 AM EDT |
GDStewart wrote:I never said he had "control", however he does have a great deal of influence in the Linux community and his approval of systemd is only encouraging Poettering to continue feeding the monster. I actually think it's quite evident that Torvalds hardly has any influence outside of the kernel, simply because whenever he says anything negative about any component of userland software, no one listens to him. It's why despite Torvalds encouraging people to first "[print] out a copy of the GNU Coding Standards, and NOT read it. Burn them" as a "great symbolic gesture" before reading the Linux Kernel Coding Style, the GNU Coding Standards are still used (possibly even primarily for software in C). It's also why despite his numerous complaints about GNOME 3 (the top bar is black? OMG!), they never fixed anything he had a problem with. GDStewart wrote:I don't care if systemd adds some replacement for su or not. I don't care if systemd sucks in every GNU utility on Earth. I just want to have a choice to use it or not use it and not have that choice severely limit which distributions I can use. I can understand this viewpoint if you aren't using SystemD. However, if you are using SystemD, even if you don't want to, you should have a problem with this. Even if the new "su" replacement is something we aren't forced to use, or worse yet, people simply never use it and continue to prefer "su", that still means more code was added for something that wasn't necessary for SystemD to include. Meaning SystemD gets bigger, more bloaty, and possibly slower. All because Poettering didn't think the 30+ year tried-and-true "su" was good enough, and didn't even think to make "su" do what he needs (perceives to need) before duplicating the functionality in SystemD. As you say, options for distributions that don't use SystemD are becoming severely limited. That's why those concerned should be the ones either telling Poettering to stop it, or going to the developers of these projects and asking or helping them to fix the issues to prevent SystemD from absorbing them. I just want to point out, Linus has nothing to do with it. That's why he isn't interested and doesn't understand the hate SystemD gets. |
750 Aug 31, 2015 10:13 AM EDT |
Err when torvalds posted a rant on G+ about networkmanager requiring root password to change the wlan, people jumped to change that behavior virtually instantly... |
AwesomeTux Aug 31, 2015 11:46 AM EDT |
750 wrote:Err when torvalds posted a rant on G+ about networkmanager requiring root password to change the wlan, people jumped to change that behavior virtually instantly... If I remember correctly, this was specific to openSUSE, and maybe Ubuntu. It was the "Make available to other users" check box not being checked by default, which isn't the normal behavior. Changing anything related to time still requires a password... Likewise with adding a printer, still behind a password. There is still a lot of "gothic" black in the GNOME theme. You still can't change font sizes without GNOME Tweak Tool. And those are just the recent things I've heard him complain about... ... more then 2 years ago. |
BernardSwiss Nov 01, 2015 12:19 AM EDT |
I just stumbled across this, today... (the internet has some funny detours, sometimes) - Busybox removes systemd support http://git.busybox.net/busybox/commit/?id=accd9eeb719916da97... |
mrider Nov 01, 2015 10:51 AM EDT |
The commit comment is priceless:Quoting:remove systemd support |
tuppp Nov 01, 2015 1:09 PM EDT |
Bravo! If only more upstream developers would likewise stand up against these systemd clowns... |
linux4567 Nov 05, 2015 8:39 AM EDT |
The only one who can stop Poettering is his employer, REDHAT. Nobody would be using Poettering 'creations' if REDHAT wasn't pushing systemd. Systemd is a REDHAT project, not a personal project of Poettering, they just like to hide behind Poettering so it doesn't seem so obvious that REDHAT is pushing systemd to gain full control over the whole Linux ecosystem. Systemd is REDHAT's trojan horse to shape and control the whole Linux ecosystem according to their own ideas. GNU/Linux is turning into REDHAT/Linux. |
jdixon Nov 05, 2015 9:25 AM EDT |
For a while now I've been seriously thinking that it might be time to create a non-business focused fork of the Linux ecosystem, aimed at hobbyists and home users. But not being a software person, that's way beyond anything I could manage. And it would almost certainly never be a money making proposition, so it would have to be a labor of love. Maybe Devuan will allow that to become possible. We'll see. |
CFWhitman Nov 05, 2015 10:30 AM EDT |
Red Hat could be looking to exert more influence in the Linux sphere. However, if they really are doing that, I think they've forgotten with whom they're dealing here. If the systemd issues ever get to the point where Red Hat can somehow leverage them to get other distributions, companies, or people to do something they don't want to do, then systemd can be dropped or forked faster than they might think. It's called Free Software for a reason. |
tuppp Nov 05, 2015 1:57 PM EDT |
CFWhitman wrote:If the systemd issues ever get to the point where Red Hat can somehow leverage them to get other distributions, companies, or people to do something they don't want to do...This scenario has already happened many times. The systemd folks have been getting developers to unnecessarily designate systemd as an external dependency. Many distros have caved since this practice started. Here's Poettering pushing for Gnome's external dependency on systemd back in May of 2011. Hopefully, more developers will renounce systemd before things get critical. |
CFWhitman Nov 05, 2015 2:34 PM EDT |
I think you're missing the degree of what I'm talking about. I'm well aware of distributions like Debian and Ubuntu changing or planning a change to systemd. However, right now it's just, 'Things are easier if we use systemd as the init.' It's not, 'We have no choice but to use systemd.' Also, there are distributions like Gentoo that have no plans to leave their default init behind to switch entirely to systemd. If it were to ever get to the point where Red Hat basically tried to force these distributions to do things the systemd way against their will, they'd revolt. Busybox is one small and important project that's already gotten fed up with systemd. If they push it too far, other projects will follow. My point is, for Free Software, there is no point of no return where you just have to accept someone else's terms, because the leverage just doesn't exist. If they wanted to, Debian could turn around and reject systemd before the next major release and go to something else altogether. It doesn't matter how deeply involved in systemd various projects get, there is always a way out. If you don't believe me, ask the folks at Xfree86. |
linux4567 Nov 05, 2015 10:02 PM EDT |
@CFWhitman: theoretically you are right, distros can reject systemd and choose something else. But in practice the cost in terms of man hours and know-how required to disentangle the fast growing systemd dependencies is already becoming so high that no smaller distro can afford to not use systemd and continue using SysV init or upstart. The only distro that had enough man power to do that long term was Debian, but they caved in, so there is no hope left for others. You need to consider that while there are countless Linux distros, most are either based on Debian or on Fedora. They do not have the manpower to create a distro from zero that's why they have to follow either Debian or Fedora. Only Slackware and Gentoo are not directly dependent on the work of Debian and Fedora and therefore might still be able to avoid systemd for a while but even for them the cost is rising due to more and more applications becoming dependent on systemd. |
BernardSwiss Nov 06, 2015 12:59 AM EDT |
If it's not a legit tactic for Microsoft, perhaps it's a dubious practice for Red Hat, as well? |
750 Nov 06, 2015 3:51 AM EDT |
Just to throw some numbers in here. Red Hat: US$ 1.534 billion in revenue for 2014. Canonical: US$65.7 million for 2013 (latest wikipedia had) Also, for the latest (4.3) kernel Red Hat was the biggest contributor besides Intel. RH is massive in the Linux ecosystem, and so what they back is very likely to become the de-facto standard. |
jdixon Nov 06, 2015 7:04 AM EDT |
> Only Slackware and Gentoo are not directly dependent on the work of Debian and Fedora and therefore might still be able to avoid systemd for a while but even for them the cost is rising due to more and more applications becoming dependent on systemd. Yes. I'm hoping Slackware can hold out or Devuan will be successful, but we'll have to see what happens. Systemd is simply too Windows-like for my tastes. |
CFWhitman Nov 06, 2015 10:59 AM EDT |
Again, the comments about Red Hat already having imposed their will on other distributions here are only talking about creating a system that a lot of people end up using because it was made the most convenient. There is a big gulf between doing that and trying to leverage it to force distributions to do what you want. As much as some people may feel that Red Hat forced Debian, Ubuntu, Arch, etc. to use systemd, they didn't really. Those projects decided to use it because they came to the conclusion that it was the better course to take. As soon as Red Hat does something onerous with systemd that they are strongly against, it will be dropped if necessary. You only have so much leeway in the Free Software world. Imagine how quickly systemd would be dropped if they introduced a license clause that was unacceptable. I don't care how big Red Hat is, an unacceptable license clause would get systemd dropped or forked fast enough to make your head spin (I don't think Red Hat is stupid enough for that yet). The thing that is really bad about systemd is that it makes it much more difficult to have alternative standards. That is, not that it would be so difficult for most distributions to drop systemd, but that it is much more difficult for there to be some distributions that use it and some that don't. Up till now, whatever init system you liked was easy enough to switch to from the most common, SysV init. Many have existed over the years, and distributions have used what they wanted. Systemd's dependency arrangement makes it so it's much more trouble to deviate from what is becoming the most popular init system. This is very similar to X-Window, where almost no distributions used anything but Xfree86. However, that still didn't stop Xfree86 from being dropped as soon as they did something that many distributions had an issue with. In reality, Xfree86 had a much stronger position than systemd can ever have, since Xfree86 was used on other Unix and Unix-like operating systems while systemd cannot be, and yet they were replaced very quickly when they made the wrong move. My guess is that eventually, systemd will either get its act together (Poettering's exit would likely be necessary), be replaced by a fork (this may be the most likely), or simply be replaced by one of the alternatives. This is not to say that they haven't or won't cause plenty of trouble in the meantime. However, I don't think systemd will give Red Hat significantly more control over the whole of Linux than they already have. There is some chance that the outcome will be that eventually there is a more general split of Linux into systemd and non-systemd based operating systems, which is probably the worst case scenario. It's also quite possible that the actions of Busybox are early indications of a general revolt which would culminate in one of the outcomes I already mentioned. |
Fettoosh Nov 06, 2015 11:33 AM EDT |
+100 That is sensible & the voice of reason. Open source corporation or not, Red Hat is doing what corporations will always try to do. Control As long as SystemD is covered by a GPL, there shouldn't be a major concern. The minute it doesn't, it should be, and will be forked. Linus would be concerned only when SystemD developers start dictating how things should be done in the Linux kernel. All good things will end when something better comes along. |
jdixon Nov 06, 2015 11:42 AM EDT |
> There is some chance that the outcome will be that eventually there is a more general split of Linux into systemd and non-systemd based operating systems, which is probably the worst case scenario. I'm not certain that would be the worst case scenario. The forced marriage of home/hobbyist and enterprise users in Linux is not necessarily a good thing. Distributions aimed at the home/hobbyist market that strip out enterprise oriented features such as systemd might be better. |
cybertao Nov 06, 2015 10:15 PM EDT |
CFWhitman wrote:It's also quite possible also that the actions of Busybox are early indications of a general revolt which would culminate in one of the outcomes I already mentioned.It could go the other way, perhaps. I'm not up to play on how initrd works with distributions using systemd but it might lead to Busybox being replaced for that purpose. In my experience systemd has proven itself for resource management on headless and larger 'embedded' setups. Busybox could potentially be forked or surpassed in favour of systemd. |
linux4567 Nov 08, 2015 9:39 AM EDT |
Busybox's choice is irrelevant in this discussion as they are not a full fat general purpose distro but rather a tiny minimalistic distro with a very narrow focus, so for them doing without systemd is far easier than for a normal distro. |
Posting in this forum is limited to members of the group: [ForumMods, SITEADMINS, MEMBERS.]
Becoming a member of LXer is easy and free. Join Us!