timcuthBraves Fan Premium Member join:2000-09-18 Pelham, AL |
timcuth
Premium Member
2014-Sep-2 12:39 pm
Poettering's view of the future of Linux |
|
ExodusYour Daddy Premium Member join:2001-11-26 Earth |
Exodus
Premium Member
2014-Sep-2 12:47 pm
Can you TLDR this? I read through most of it, but maybe I'm just not getting it. They want to get rid of package managers and use what instead? |
|
SeleniaGentoo Convert Premium Member join:2006-09-22 Fort Smith, AR
1 recommendation |
to timcuth
Probably Ubuntu users don't care, but it seems what they are discussing will thwart shared libraries, with each app running in its own root, similar to chroot in a way, but in its one of many runtimes for the same system.
I agree this may be useful for troublesome apps to use an exact environment it was developed under, but I do not think it should be default, rather an option that can be enabled per app. I can just picture this 7 different versions of the same lib running for different apps and that is assuming that the same version can be reused, which they did not specify.
Otherwise, it would be worse. We all think Ubuntu and Windows(especially) are bloated? This would make Ubuntu and Windows seem lean in comparison, if I am getting what they are trying to do. I say let them work because they may eventually work out these issues as it is an idea in its infancy. If they don't, the nice thing about FOSS is the community chooses to adopt something or not adopt it. It is not forced on you like with Windows 8 Metro from Microshaft. |
|
timcuthBraves Fan Premium Member join:2000-09-18 Pelham, AL Technicolor ET2251
1 recommendation |
timcuth
Premium Member
2014-Sep-2 2:48 pm
I suppose my big thought is, why don't they create some new OS that works this way, instead of subsuming all (or most) Linux distros into their grand plan? My personal statement of the tldr; is that systemd will become the major part of the system, and "distros" will be relegated to mountable subvolumes which are essentially the /usr directory of Linux systems. Linux will no longer be an OS, it will simply be (multiple) add-ons to the systemd system. And, imho, that sucks. Tim |
|
2 recommendations |
said by timcuth:My personal statement of the tldr; is that systemd will become the major part of the system, and "distros" will be relegated to mountable subvolumes which are essentially the /usr directory of Linux systems. Linux will no longer be an OS, it will simply be (multiple) add-ons to the systemd system. It's all part of Lennart Poettering's diabolical scheme to rule the universe Count me as a "systemd" skeptic. It seems to be an invasive cancer. I am not a member of the Poettering fan club. |
|
2 recommendations |
to timcuth
Seems to be the same thing systemd was, a (poor) solution looking for a problem. This on the surface appears to be a management/configuration/bloat nightmare. |
|
|
rchandraStargate Universe fan Premium Member join:2000-11-09 14225-2105 ARRIS ONT1000GJ4 EnGenius EAP1250
3 recommendations |
to timcuth
Let's just say, systemd is evil, and be done with it.
It breaks the Unix philosophy of doing one thing and doing it well, with interfaces which make it easy to put data in and get data out. Instead, it's a metasticizing cancer which wants to gobble up init, udev, cron, dbus, and what--now FS mounts???
The developers show a dumbfounding lack of community, when they started using the long-standing "debug" keyword for their operations, and basically said to the community, screw you, we don't care what has been used in the past, this is the way we're going to use it. Someone even proposed a kernel patch which would deny systemd access to the kernel command line due to their intransigence.
What they seem to be proposing can already be covered, mostly. When a package needs some functionality, it declares a dependency on something else, possibly including version information...e.g., openssl098 vs openssl100 or somesuch. Through careful naming and such, multiple versions can coexist. Where for some reason they can't, it's possible to do "interesting" things with LD_LIBRARY_PATH and such.
I don't even think their solution is particularly relevant to the problem they say they're trying to solve: having a developer's work be migrated easily to several different systems. It seems to me the developer's work will only be properly built and tested on either discreet systems or some collection of virtualized machines. My armchair assessment of it is, all this mucking about with what's mounted in what namespace is just going to generate more trouble than it's worth. I do understand somewhat what they're trying to do, that is, having only single copies of the relevant parts of systems around, and bolt on variable stuff to that, but it seems kind of unreasonable for all end-users' systems to have all this extra machinery, just because systemd wants to be the central point in all of that. |
|
SeleniaGentoo Convert Premium Member join:2006-09-22 Fort Smith, AR 1 edit |
to timcuth
Is the sky really falling? » boycottsystemd.org/To be fair, let's let one of the systemd developers have his say. » 0pointer.de/blog/project ··· why.htmlI will say for my use so far, I notice no issues. In fact, I did not even notice when Debian made the switch right away until I saw systemd in my processes a little while after upgrading to Jessie and looked up what it was. |
|
|
said by Selenia:Is the sky really falling? »boycottsystemd.org/
To be fair, let's let one of the systemd developers have his say. »0pointer.de/blog/project ··· why.html
I will say for my use so far, I notice no issues. In fact, I did not even notice when Debian made the switch right away until I saw systemd in my processes a little while after upgrading to Jessie and looked up what it was. I tried it out for a while on a Gentoo build, can't say that I liked it much. For the most part everything just worked, however I could never get my networking to work properly. For everything I tried I could never get the wireless card to connect. So I went back to OpenRC and all is well. As for what the devs have to say, really highlighting a project green because it uses git over svn or Bazaar tells me what I need to know. Don't get me wrong, I like git and predominately use it myself, but to say that using it is green while other equally as good SCM options are red is telling to me. |
|
reub2000 Premium Member join:2001-12-28 Evanston, IL |
to timcuth
It seems like a nice project. It just seems like that it has a bit of mission creep. It should stick to starting and managing deamons. |
|
SeleniaGentoo Convert Premium Member join:2006-09-22 Fort Smith, AR |
to GILXA1226
For me, it worked better on x86 and x86_64 for detecting and initializing hardware. ARM is a whole different scenario. I had to manually roll back to upstart for that and pin the necessary dependencies. It is a young project, so I can cut them slack there. I don't have anything against a uniform framework to help devs make things work, so long as they can bolt on anything they need and use more suitable substitutes for a certain function, as needed. I do not see this being impeded thus far. |
|
|
to timcuth
I heard that systemd will be developing their own kernel and no longer need Linux. |
|
SeleniaGentoo Convert Premium Member join:2006-09-22 Fort Smith, AR
1 recommendation |
Selenia
Premium Member
2014-Sep-2 8:38 pm
That is when the Linux community will shift to something else. That is the beauty of open source, the community decides which tools are useful for the time being and are free to change them up, based on users' and their own needs. Right now, systemd seems to be useful to developers. If it turns out otherwise, it will either be forked or something else will be dropped in its place. Does not mean we should discourage a project that aims to encompass a lot of functionality from being developed. Ultimately, the community will decide whether it aids in their mission or something else useful may fork from it. |
|
1 recommendation |
to Exodus
said by Exodus:Can you TLDR this? I read through most of it, but maybe I'm just not getting it. They want to get rid of package managers and use what instead? My understanding is that it would use btrfs snapshots in a unique way. When you go to distribute software, instead of just making a git tree and letting the distro handle it, you could bundle it and all dependencies it needs into a btrfs snapshot. You then sign that shapsnot and distribute it. On the user end, they receive that snapshot, copy it to their drive and then installation is complete. The snapshot gets deduped if it contains libraries the system already contains and you are then good to go. From a support standpoint, it simplifies things greatly. You no longer have to wrangle dependencies and the distro a user is running would be irrelevant. You can know that everything is present and that everything is accurate and untampered with (due to checksumming and the signing). Updates are also smaller as a btrfs send/recv does an rsync style download to only download changed files. The rest of the article relates to how this entire concept could apply to the entire OS and how it would all be organized and such. It is a pretty unique concept and would allow for some very cool things, but as an implementable project, it is still at least several years away from a functional prototype. |
|
dave Premium Member join:2000-05-04 not in ohio
1 recommendation |
to timcuth
said by timcuth:I suppose my big thought is, why don't they create some new OS that works this way, instead of subsuming all (or most) Linux distros into their grand plan? So what forces these distros to go along with the plan? Presumably they think it's an improvement over what they had before. Given that they freely and openly choose what to do with their free and open software, there doesn't seem to be much to object to on grounds of principle. Of course, you're entirely entitled to prefer a different solution, but that's a separate issue. |
|
3 recommendations |
said by dave:So what forces these distros to go along with the plan? Many distros bought in to systemd when it was an init system (see the prolonged Debian voting process). Now that they're stuck with it for the next couple releases, it turns out that systemd wants to be an operating system or distribution unto itself. Not great. |
|
2 recommendations |
Salty_Peaks to timcuth
Anon
2014-Sep-3 6:10 pm
to timcuth
I don't like systemd. I hated Upstart and all the race conditions. We are reaching an inflection point where what is good for the desktop is not good for the server. I would love to bring back System V where at least things were executed in an intelligent order and you didn't have weird semaphores and parallelization issues that these new start-up daemons bring.
What confuses me the most is that these desktop-centric technologies continue to find their way to server systems which more oft than not only reboot to effect glibc/openssl/kernel updates.
This whole systemd fiasco smells like Unity. |
|
SeleniaGentoo Convert Premium Member join:2006-09-22 Fort Smith, AR
2 recommendations |
Selenia
Premium Member
2014-Sep-3 7:53 pm
But hey, systemd did solve the race issues. Have not seen 1 yet. Of course, SysVinit never had that issue either. Every solution has its own problems, which means devs and users choosing the one that won't get in their way.
Oh and 1 difference: Systemd has not yet got in my way. Unity, on the other hand, constantly got in my way within 5 minutes. #effubuntu |
|
1 recommendation |
to timcuth
Well.. in a word or two...
No thanks...
Why? Well for starters... "This is a reinvention...." BZZTT!!! NO THANKS!
Any time I see "reinvention" or re-imagine or reboot.. it is like all the movies and TV shows they "reinvent." GARBAGE....
Second.. this EFI SecureBoot BS, again....I don't care how you spin this... yes it may have some good and desirability in a FEW SMALL NICHE AREAS.. but the *nix/BSD chain is already FAR MORE SECURE to start than where this whole BS started.. NO! NO and JUST SAY NO! You are right.. I want nothing to do with this mess... I know ultimately where all this is headed.. and Ubuntu for one has stated they are not going to play this game... so...
Third, this falls into the same position that another project does... a solution looking for a problem to fix.
Lastly. the key part of this, and thus its ultmitate still birth is: ..."hence needs great support from the distributions." Yeah.. good luck with that! We can get the distros to agree that DEB is the superior and best packaging method and ultimately kill off the garbage RPM.. and you think this is going to happen....HAHAHAHA..
Summary: No thanks. The future of Linux would be a lot better off if we concentrated all of this wasted effort into things we don't need to the things we DO NEED. This is not one of them. |
|
|
said by TuxRaiderPen2:Third, this falls into the same position that another project does... a solution looking for a problem to fix. It seems to me that the "problem" it's intended to fix is, "It's too hard to distribute closed-source software for Linux." To which, I can only say: Fuck you, that's a feature. |
|
1 recommendation |
to timcuth
I can only say that as I migrate to newer releases of my preferred distro in my job, I spend more time fixing things broken by systemd than most anything else. It's always something. |
|
2 recommendations |
to Salty_Peaks
My final thoughts here -- we're reacting to this change and effort as-if we're victims, we're not. The nature of FOSS dictates that if we don't like it fork()it and I'm done. Ubuntu didn't just erupt from chaos, it manifested due to a need. The same for SuSE, CentOS (FU and you're "For the community, not by it" hilarity), Scientific Linux, et al.
C'mon, we're all Open Source, we dont' like it lets change it. Distro "wars" fleshes out ideals versus value. Unity is great, Gnome3 is awesome... oh hello MATE. |
|
SeleniaGentoo Convert Premium Member join:2006-09-22 Fort Smith, AR
2 recommendations |
Selenia
Premium Member
2014-Sep-8 11:30 pm
True, I can't stand Unity and am not much a fan of GNOME 3, but I love KDE. I can install KDE on any distro, no matter what came as the default and some distros give you a choice up front in the install process. I use XFCE in my ARM based experiments as they mainly function as servers only needing a desktop for a few utilities I use dependent on a display(I VNC in for those). Bottom line of FOSS is being able to customize for your needs and wants. Developers do the same to accomplish the mission they are trying to accomplish with what they develop. I would not worry. In fact, I think systemd could provide a good model for standardizing some things, even if it becomes overbearing, which would then give rise to a fork. |
|
SteveI know your IP address
join:2001-03-10 Tustin, CA |
to TuxRaiderPen2
Huh? Where does this come from? |
|
dave Premium Member join:2000-05-04 not in ohio |
dave
Premium Member
2014-Sep-9 4:57 pm
Conspiracy theory fans known that everything is secretly connected! |
|
SeleniaGentoo Convert Premium Member join:2006-09-22 Fort Smith, AR |
Selenia
Premium Member
2014-Sep-9 8:34 pm
I think he misunderstood the part where one goal for systemd was a uniform platform for handling UEFI Secure Boot, as you still have to hack many distros who refuse to buy a key from MS to get it working with Secure Boot enabled, which is proving a problem for some OEM Windows 8/Linux dual booters, from what I been reading. I don't dual boot so no issue thus far. |
|
1 recommendation |
to timcuth
Sounds like Poettering is railroading Torvalds. Eventually Poettering would like to be king. But many lives will be lost in the process and much destruction.
666 |
|